> ## Documentation Index
> Fetch the complete documentation index at: https://slng.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# SLNG changelog

> Release notes for the SLNG speech and language API. New TTS and STT models, voice agent features, breaking changes, and bug fixes.

**September 3, 2026**

API v0.69.6

## Retired language variants for Nova 3, Aura 2, and Arcana

Six language-specific endpoints are removed from the public API: Deepgram Nova 3 Multi-Language speech-to-text, Deepgram Aura 2 Spanish, Rime Arcana v2 Arabic, and Rime Arcana v3 French, Hindi, and Spanish. Remaining variants are unchanged, including [Nova 3](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-http.md), [Aura 2 English](https://slng.mintlify.site/api-reference/tts/deepgram-aura-2/aura-2-english-http.md), [Arcana v2 Spanish](https://slng.mintlify.site/api-reference/tts/rime-arcana-v2/arcana-v2-spanish-http.md), and [Arcana v3 English](https://slng.mintlify.site/api-reference/tts/rime-arcana-v3/arcana-v3-english-http.md). Check the [model catalog](https://slng.mintlify.site/models/index.md) for current alternatives per language.

**September 1, 2026**

API v0.69.5

## Fish TTS 2.1 Pro on SLNG infrastructure

[Fish TTS 2.1 Pro](https://slng.mintlify.site/api-reference/tts/fish-tts-2-1-pro/fish-tts-2-1-pro-http.md) is now hosted on SLNG infrastructure, serving from EU and AU regions over HTTP and [WebSocket](https://slng.mintlify.site/api-reference/tts/fish-tts-2-1-pro/fish-tts-2-1-pro-ws.md). You get 58 languages, 280 ms measured latency, instant voice cloning from 10 seconds of audio, and 60+ emotion tags at \$0.011/min.

## Fish Audio voice library

A new [Fish Audio voices page](https://slng.mintlify.site/voices/fish-audio.md) lists 338 voices across 59 languages with audio samples. Pass a voice's `voice_id` as the `reference_id` field in your request, or leave it unset for the default voice.

## 30 more languages in the language catalog

The [language catalog](https://slng.mintlify.site/models/by-language.md) adds 30 languages, including Kazakh, Kurdish, Lao, Macedonian, Nepali, Serbian, Somali, and Zulu. Coverage comes from expanded language support in Inworld Max 1.5, Soniox TTS RT v1, Deepgram Nova 3, Speech AI RT v5, and Kugel 2.

## Cleaner tool listings in the shared tools API

Listing tools now returns only your organisation's live tools. Curated SLNG tools no longer appear mixed into the list. Fetching a single Custom Code tool now includes its source code, dependencies, declared secrets, and introspected argument schema. MCP server responses now include each tool's input schema and a schema hash so you can detect contract changes.

**August 27, 2026**

API v0.64.0

## Shared tools for voice agents

Agent tools now live in a shared organisation library. You create a tool once, test it, publish an immutable version, and attach that version to any number of agents. Publishing a new version never changes a live agent until you upgrade its attachment. The library supports [Custom Code tools](https://slng.mintlify.site/examples/agents-config.md#tools) that run typed Python in an offline sandbox, API Request tools with Vault-backed authentication, capability tools such as End Call and Transfer Call, and remote MCP servers.

## Agent resources API reference

A new [Agent resources](https://slng.mintlify.site/api-reference/overview.md) section documents the endpoints for creating, testing, publishing, and attaching tools, connecting MCP servers, managing [Vault](https://slng.mintlify.site/dashboard/vault.md) secrets, and registering client models.

## Measured latency in the model catalog

The [model catalog](https://slng.mintlify.site/models/index.md) now shows measured latency in milliseconds for each model, such as 109 ms for Cartesia Sonic 3.5, replacing the previous low/medium labels. Models without a measurement show `n/a`.

## 24 more languages in the language catalog

The [language catalog](https://slng.mintlify.site/models/by-language.md) adds 24 languages, including Hungarian, Indonesian, Persian, Swahili, Ukrainian, Urdu, and Welsh. Coverage comes from Speechmatics Batch transcription and Fish Audio TTS. Use it to find every model that supports your target language.

**August 25, 2026**

API v0.63.5

## Speechmatics Realtime v2 STT

[Speechmatics Realtime v2](https://slng.mintlify.site/api-reference/stt/speechmatics-realtime-v2/speechmatics-realtime-v2-ws.md) is a new SLNG-hosted speech-to-text model over WebSocket. Stream binary audio and receive partial and final transcripts with word timing, confidence, and optional diarization. You can also force end of utterance to finalize a transcript without closing the stream. A dedicated [Indonesian variant](https://slng.mintlify.site/api-reference/stt/speechmatics-realtime-v2/speechmatics-realtime-v2-indonesian-ws.md) is available alongside the multilingual endpoint.

## Jakarta deployment region

The [regional model catalog](https://slng.mintlify.site/models/by-region.md) now lists **Jakarta (asia-southeast2)**, serving [Deepgram Nova 3 Indonesian](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-indonesian-ws.md) and Speechmatics Realtime v2. Use it to run Indonesian speech-to-text close to Southeast Asian traffic.

## Eemshaven replaces EU North in the catalog

The European region **EU North (eu-north-1)** is replaced by **Eemshaven (europe-west4)** in the [regional model catalog](https://slng.mintlify.site/models/by-region.md). Speechmatics Batch transcription is served from Eemshaven. Check the region tabs if you pinned models to eu-north-1.

**August 19, 2026**

API v0.60.2

## Fish Audio TTS and ASR

Fish Audio joins the model catalog with two TTS models — [Fish Audio TTS S2 Pro](https://slng.mintlify.site/api-reference/tts/fish-audio-tts-s2-pro/fish-audio-tts-s2-pro-ws.md) and [Fish Audio TTS S2.1 Pro](https://slng.mintlify.site/api-reference/tts/fish-audio-tts-s2-1-pro/fish-audio-tts-s2-1-pro-ws.md) — plus the [Fish Audio ASR](https://slng.mintlify.site/api-reference/stt/fish-audio-asr/fish-audio-asr-http.md) speech-to-text endpoint. Both TTS models expose HTTP and WebSocket transports.

## Cartesia Sonic 3.5 TTS

[Cartesia Sonic 3.5](https://slng.mintlify.site/api-reference/tts/cartesia-sonic-3-5/cartesia-sonic-3-5-ws.md) is available over WebSocket in the Cartesia hosting tier. Pick it alongside Sonic 3 from the Cartesia model group.

## New deployment regions

The [regional model catalog](https://slng.mintlify.site/models/by-region.md) now lists **US Central (us-central1)** in North America and **Mäntsälä (nebius-eu-north1)** in Europe, plus cross-region availability sections for Asia-Pacific and Middle East & Africa. Use the region tabs to see which models run closest to your traffic.

## Batch API authentication and workflow refresh

The [Batch API guide](https://slng.mintlify.site/batch-guide.md) now documents the `slng_cu_` (consumer) and `slng_bt_` (batch) key formats, and both the `Authorization: Bearer` and `X-API-Key` headers work equivalently. Multipart file upload is retired — use URL input, presigned S3 upload, or the new inline base64 method for small files. New sections cover the `DELETE /v1/batch/jobs/{job_id}` endpoint, the full error status table, and an end-to-end presigned-upload workflow example.

## Batch API limits updated

Request bodies are now capped at **10 MB** (use presigned upload for larger files) and presigned upload URLs expire after **1 hour** instead of 30 minutes. The default `operating_point` is now `enhanced` rather than `standard`.

**August 13, 2026**

API v0.58.1

## Vault for organization-wide variables and secrets

The [Vault](https://slng.mintlify.site/dashboard/vault.md) is a new Dashboard area for values your agents share. Store reusable strings as **Variables** and write-only credentials as **Secrets**, then reference either as `{{$NAME}}` in prompts, greetings, and templated tool fields. Both kinds are encrypted at rest and scoped to your organization.

## Bring Your Own Key for managed agents

[Bring Your Own Key](https://slng.mintlify.site/dashboard/byok.md) lets managed voice agents run against your own LLM and speech-model provider accounts. Register OpenAI-compatible, Azure OpenAI, or Google Vertex AI endpoints, plus STT and TTS models from the SLNG catalog, and pick them from any agent's model dropdown. Billing and rate limits move to your provider contract.

## Version history, export, and import for agents

Every save now appends an immutable version to an agent's history, and restoring a version is a preview-then-accept flow so you can review field changes before applying them. `GET /v1/agents/{agent_id}/config` downloads a portable JSON config (secrets excluded) and `POST /v1/agents/import` recreates the agent from one. See [Version history, export, import, and duplication](https://slng.mintlify.site/examples/agents-api.md) for the full contract.

## Realtime data messages for the Web embed

The [Web embed guide](https://slng.mintlify.site/agents/embed-web.md) now documents every realtime data message the agent room emits, including `agent_state_changed` for driving listening/thinking/speaking UI without inferring from audio levels. The same state is mirrored to the `lk.agent.state` participant attribute.

## LiveKit plugin guide updated for 1.6.10

The [LiveKit plugin guide](https://slng.mintlify.site/agents/livekit-plugin.md) is refreshed for the released `livekit-plugins-slng` on `livekit-agents>=1.6.10`. New coverage: `extra_headers` and `runtime_init`, the deterministic 4xx failover rule, a Soniox rt-v5 failover example, and a warning that language codes are sent verbatim (Sarvam Bulbul needs BCP-47 like `hi-IN`, not `hi`). Upgrading from 1.6.6 or earlier: `model_endpoint` and `model_endpoints` were removed and now raise — pass `model=` or `connections=` instead.

## Agents OpenAPI spec caught up

The Managed Agents API reference now matches the live API, including the version-history, config export/import, and duplicate endpoints above.

**July 29, 2026**

API v0.57.7

## Context Router for OpenAI-compatible voice agents

The [Context Router](https://slng.mintlify.site/context-router.md) is a new OpenAI-compatible chat endpoint that adds response caching in front of your existing LLM without changing agent logic. Send a normal Chat Completions request with model `slng/auto`, and SLNG picks the underlying model and serves cached answers when it can. Region-specific base URLs are available for India, US, Indonesia, and EU.

## ElevenLabs Conversational AI integration

Point an ElevenLabs Conversational AI agent at the Context Router as its **Custom LLM** to keep your voice, prompt, and tools while gaining response caching. See [Connect ElevenLabs to the Context Router](https://slng.mintlify.site/llm-router/elevenlabs.md) for the dashboard and API setup.

## Vapi assistant integration

Route a [Vapi](https://slng.mintlify.site/llm-router/vapi.md) assistant through the Context Router by setting a custom OpenAI-compatible provider URL on its model config. Your assistant keeps working unchanged and picks up router-side caching with a base-URL swap.

**July 22, 2026**

API v0.57.6

## Full Bengali voice roster for Sarvam Bulbul

[Sarvam Bulbul](https://slng.mintlify.site/voices/sarvam-bulbul.md) now ships a complete 37-voice Bengali (`bn-IN`) lineup, up from a placeholder set. Pass any voice ID — like `ishita`, `kabir`, or `priya` — with `language: "bn-IN"` for Bengali synthesis.

## Kannada and Tamil rosters trimmed on Sarvam Bulbul

Kannada (`kn-IN`) and Tamil (`ta-IN`) on [Sarvam Bulbul](https://slng.mintlify.site/voices/sarvam-bulbul.md) each drop from 37 to 36 voices. If you were pinning `mohit` for Kannada, switch to another masculine voice from the updated list.

## Slovenian voice list cleaned up on Soniox

Slovenian (`sl`) on [Soniox](https://slng.mintlify.site/voices/soniox.md) now lists 4 voices instead of 15. The removed entries were duplicate and unsupported IDs; the four remaining voices are the ones that render correctly.

**July 15, 2026**

API v0.57.5

## Hindi voices for Cartesia Sonic 3

[Cartesia Sonic 3](https://slng.mintlify.site/voices/cartesia-sonic-3.md) now offers 42 Hindi voices covering conversational, narration, customer support, and Hinglish use cases. Pass any of the new voice IDs — like `Aadhya`, `Ashwin`, or `Meera` — with `language: "hi"` for Hindi output.

## More Bengali voices for Cartesia Sonic 3

Bengali on [Cartesia Sonic 3](https://slng.mintlify.site/voices/cartesia-sonic-3.md) grew from 2 to 3 voices. Use `language: "bn"` to synthesize Bengali speech with the expanded roster.

**July 6, 2026**

API v0.57.4

## Inline base64 audio submission for Batch API

You can now submit small audio files to the [Batch API](https://slng.mintlify.site/batch-guide.md) as base64 in the request body — no upload step, no public URL. Pass `input_base64` with `input_filename` alongside your `transcription_config`. Use it for clips under the 10 MB request-body cap; larger files still need presigned upload.

## Delete Batch API jobs

Call `DELETE /v1/batch/jobs/{jobId}` to remove a completed or failed job and its S3 inputs and outputs. The endpoint returns `204` on success and `409` if the job is not yet in a terminal state. See the [Batch guide](https://slng.mintlify.site/batch-guide.md).

## Richer transcription configs on Batch jobs

Batch job submissions now accept `tracking`, `output_config`, `translation_config`, `summarization_config`, `sentiment_analysis_config`, `topic_detection_config`, `auto_chapters_config`, and `audio_events_config`. Attach titles, tags, translations, summaries, sentiment, topics, chapters, and audio-event detection at job creation. See the [Batch guide](https://slng.mintlify.site/batch-guide.md) for the field list.

## Batch API default accuracy is now `enhanced`

The default `operating_point` for Batch transcription changed from `standard` to `enhanced`. New jobs pick up the higher-accuracy model automatically — set `operating_point: "standard"` explicitly if you want the previous behavior.

## Multipart file upload retired on Batch API

The `multipart/form-data` upload path (formerly Method 1) has been removed from the [Batch API](https://slng.mintlify.site/batch-guide.md). Submit audio via URL input, presigned S3 upload, or the new inline base64 field instead.

## Only `slng_*` keys authenticate the Batch API

The Batch API now accepts only keys of type `batch` created at [app.slng.ai/api-keys](https://app.slng.ai/api-keys). Consumer keys and any non-`slng_*` value return `401`. Send the key as `Authorization: Bearer …` or `X-API-Key: …` — both headers are equivalent.

**July 1, 2026**

API v0.57.3

## Refreshed Soniox TTS v1 voice catalog

The [Soniox TTS v1](https://slng.mintlify.site/voices/soniox.md) voice list now reflects current availability. Afrikaans keeps its full 12-voice lineup, and Slovenian is available with `Maya`, `Daniel`, `Noah`, and `Nina`. Other languages previously listed on the page are no longer offered — update any hardcoded `voice` values that referenced them.

## Refreshed Sarvam Bulbul voice catalog

The [Sarvam Bulbul](https://slng.mintlify.site/voices/sarvam-bulbul.md) page now shows the current per-language voice counts across Bengali, Gujarati, Kannada, Malayalam, Marathi, Odia, Punjabi, Tamil, and Telugu. English, Hindi, and the standalone `hi` group have been removed from the listing.

**June 29, 2026**

API v0.57.2

## n8n community node

Run SLNG text-to-speech, speech-to-text, and outbound voice-agent calls from inside n8n workflows with the new [`n8n-nodes-slng`](https://slng.mintlify.site/sdks/n8n) community node. It ships an **SLNG** action node and an **SLNG Trigger** node so voice agents can call workflows as tools mid-call, both backed by a single API-key credential.

## LiveKit plugin no longer needs `region_override`

The `region_override` argument on `slng.STT` and `slng.TTS` is gone from the [LiveKit plugin](https://slng.mintlify.site/agents/livekit-plugin.md). Routing now happens automatically on the platform, so you can drop the parameter from existing agents and rely on the default region selection.

**June 19, 2026**

API v0.57.1

## Orpheus (English) TTS with emotion control

[Orpheus (English)](https://slng.mintlify.site/api-reference/tts/orpheus-english/orpheus-english-ws.md) by Canopy Labs is now SLNG-hosted in `eu-north-1` and `us-east-1`. Pick from eight voices (`tara`, `leah`, `jess`, `leo`, `dan`, `mia`, `zac`, `zoe`) and steer delivery with the `emotion` parameter — `neutral`, `happy`, `sad`, `angry`, or `surprised`. Stream over the unified TTS WebSocket protocol or call the [HTTP endpoint](https://slng.mintlify.site/api-reference/tts/orpheus-english/orpheus-english-http.md) for one-shot synthesis.

**June 18, 2026**

API v0.55.0

## Catalog API for model discovery

The new [Catalog API](https://slng.mintlify.site/api-reference/overview.md) exposes `GET /v1/catalog/models` and `GET /v1/catalog/models/{model_code}` so you can list and inspect every model on SLNG from your own code. Filter by service type, provider, region, language, use case, capability, or protocol, search by keyword, and page through results. Authentication is optional — pass an API key to personalize pricing and region availability for your organization.

## Account endpoint

`GET /v1/me` returns the account behind the API key making the request, including `name`, `email`, `org_id`, `org_name`, `api_key_label`, and `tier`. Use it to confirm which workspace a key belongs to before running automation. See the [Account API reference](https://slng.mintlify.site/api-reference/overview.md).

## Inworld Max 1.5 TTS

[Inworld Max 1.5](https://slng.mintlify.site/api-reference/tts/inworld-max-1-5/inworld-max-1-5-ws.md) is now SLNG-hosted for low-latency, expressive multilingual synthesis across 15 languages. Stream over the unified TTS WebSocket protocol or call the [HTTP endpoint](https://slng.mintlify.site/api-reference/tts/inworld-max-1-5/inworld-max-1-5-http.md) for one-shot synthesis. The model runs in the `nebius-eu-north1` region.

## Expanded regions for Rime Arcana v3 and Deepgram Nova 3

[Rime Arcana v3](https://slng.mintlify.site/api-reference/tts/rime-arcana-v3/arcana-v3-english-ws.md) is now available in `asia-south1` alongside `ap-southeast-2` and `eu-north-1`. SLNG-hosted [Deepgram Nova 3](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-multi-language-ws) adds `asia-south1` and `us-central1`, giving you lower-latency routing for Indian and US East workloads. Region codes for direct-proxy Deepgram, Cartesia, Murf, Sarvam, and Soniox models have been normalized — for example `na` is now `us`, and Murf Falcon now exposes per-country codes like `gb`, `jp`, `kr`, and `ae`.

**June 12, 2026**

pipecat-slng 0.4.0

## BYOK in the Pipecat plugin

[`pipecat-slng`](https://slng.mintlify.site/agents/pipecat-plugin) 0.4.0 adds Bring Your Own Key: pass your upstream provider key via the new `provider_key` argument on `SlngSTTService`, `SlngTTSService`, and `SlngHttpTTSService`, and it is forwarded as the `X-Slng-Provider-Key` header so the provider bills your account directly — no SLNG audio-minute fees. BYOK requires external catalog routes (model strings without the `slng/` prefix, such as `deepgram/nova:3`). See the plugin's [BYOK section](https://slng.mintlify.site/agents/pipecat-plugin.md#bring-your-own-key-byok) and the [BYOK guide](https://slng.mintlify.site/execution-layer/byok.md).

**June 16, 2026**

API v0.53.0

## Gradium STT and TTS

[Gradium STT](https://slng.mintlify.site/api-reference/stt/gradium-stt/gradium-stt-default-ws.md) streams real-time multilingual transcription over WebSocket, with `Flush` and `EndOfStream` controls for finalizing buffered audio. [Gradium TTS](https://slng.mintlify.site/api-reference/tts/gradium-tts/gradium-tts-default-ws.md) ships alongside it for streaming WebSocket synthesis and one-shot [HTTP synthesis](https://slng.mintlify.site/api-reference/tts/gradium-tts/gradium-tts-default-http.md).

## Kugel 2 Turbo TTS

[Kugel 2 Turbo](https://slng.mintlify.site/api-reference/tts/kugel-2-turbo/kugel-2-turbo-ws.md) is the new low-latency tier in the Kugel TTS family. Stream high-quality synthesis with expressiveness controls over the unified TTS WebSocket protocol, with word timestamps and per-chunk completion events.

## Rime Arcana v3 French

[Rime Arcana v3 French](https://slng.mintlify.site/api-reference/tts/rime-arcana-v3/arcana-v3-french-ws) joins the Arcana v3 lineup. Synthesize French speech via WebSocket or [HTTP](https://slng.mintlify.site/api-reference/tts/rime-arcana-v3/arcana-v3-french-http) using speakers like `livet_aurelie`.

## ElevenLabs WebSocket controls

The ElevenLabs TTS WebSocket endpoints, including [Flash v2.5](https://slng.mintlify.site/api-reference/tts/eleven-flash-v2-5/elevenlabs-flash-v2-5-ws.md), [v3](https://slng.mintlify.site/api-reference/tts/eleven-v3/elevenlabs-v3-ws.md), and [Multilingual v2](https://slng.mintlify.site/api-reference/tts/multilingual-v2/elevenlabs-multilingual-v2-ws.md), accept new init parameters: `auto_mode` to cut latency by skipping chunk scheduling, `sync_alignment` to attach alignment data to each audio chunk, `enable_ssml_parsing` for inline SSML, `inactivity_timeout` (1–180 s), and `chunk_length_schedule`. Set `enable_logging: false` for zero-retention mode on enterprise plans on both WebSocket and HTTP endpoints.

**June 12, 2026**

API v0.52.3

## Execution Layer documentation

The docs portal is reframed around the [Execution Layer](https://slng.mintlify.site/execution-layer.md), the three-stage path that sits between your orchestrator and the models. Read how [STT Routing](https://slng.mintlify.site/execution-layer/stt-routing), [Tiered Decisioning](https://slng.mintlify.site/execution-layer/tiered-decisioning), and [Output Assembly](https://slng.mintlify.site/execution-layer/output-assembly) cut latency and model cost across a voice call. The new [How It Works](https://slng.mintlify.site/execution-layer/how-it-works.md) page walks through a single turn end to end.

## Streamlined navigation and a single API Reference

The top bar collapses to **Documentation**, **Execution Layer**, **API Reference**, and **Models**. Every endpoint now lives under one [API Reference](https://slng.mintlify.site/api-reference/overview.md), grouped by Unified API, TTS, STT, Voice Agents, Batch, and Bridges, so you can find a model without switching tabs.

## Consolidated authentication and BYOK pages

API key creation, rotation, and request authentication are merged into a single [Authentication & API Keys](https://slng.mintlify.site/authentication.md) page. Bring-your-own-key setup for every supported provider now lives on one [BYOK](https://slng.mintlify.site/execution-layer/byok.md) page covering caching behavior and provider key handling.

**June 11, 2026**

## Pipecat plugin

The [`pipecat-slng`](https://slng.mintlify.site/agents/pipecat-plugin) Python package connects a [Pipecat](https://github.com/pipecat-ai/pipecat) pipeline to any STT or TTS model on the SLNG gateway. It ships `SlngSTTService` and `SlngTTSService` for low-latency streaming over WebSocket, plus `SlngHttpTTSService` for non-streaming synthesis. Swap the `model` string to switch provider or region without changing your pipeline code.

**June 10, 2026**

API v0.52.2

## New Sarvam Bulbul and Soniox voices

[Sarvam Bulbul v3](https://slng.mintlify.site/api-reference/tts/sarvam-ai-bulbul-v3/bulbul-v3-ws.md) adds the Tamil voice `ta-IN-rahul` and the multilingual `shubh`, while `gu-IN-rahul` is retired. [Soniox TTS v1](https://slng.mintlify.site/api-reference/tts/soniox-tts-v1/soniox-tts-v1-ws.md) adds four English voices: `Maya`, `Daniel`, `Noah`, and `Nina`.

## Updated Cartesia Sonic 3 voice metadata

The 745-voice [Cartesia Sonic 3](https://slng.mintlify.site/api-reference/tts/cartesia-sonic-3/cartesia-sonic-3-ws.md) catalog has refreshed metadata. Voice IDs are unchanged, so existing requests keep working.

**June 3, 2026**

API v0.52.1

## Gateway spec refinements

Patch-level updates to the unified gateway specification clean up schemas across the TTS, STT, and bridge surfaces. Existing requests and responses keep working without changes.

**June 2, 2026**

API v0.52.0

## Inworld Max 1.5 TTS

[Inworld Max 1.5](https://slng.mintlify.site/api-reference/tts/inworld-max-1-5/inworld-max-1-5-ws.md) is available on the SLNG-hosted TTS API in `us-east-1`. Synthesize up to 2,000 characters per request across 16 languages, pick from 100+ named voices, and tune output with `temperature`, `language` (BCP-47), and word- or character-level timestamps.

## Asia South region for Nova 3 English and Rime Arcana v3

[Deepgram Nova 3 English](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-english-ws.md) and [Rime Arcana v3 English](https://slng.mintlify.site/api-reference/tts/rime-arcana-v3/arcana-v3-english-ws.md) can now route to `asia-south1`. Set `X-Region-Override: asia-south1` to keep traffic in-region for users in India.

## Region rename for Nova 3 Indian languages

The Hindi, Marathi, Kannada, Tamil, and Telugu variants of [Deepgram Nova 3](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-ws.md) now expose a single `asia-south1` region. The legacy `ap-south-1` value is no longer accepted on `X-Region-Override`; switch any pinned requests to `asia-south1`.

**May 29, 2026**

API v0.51.2

## Supported LLMs for Voice Agents

The `llm` field on [Voice Agents](https://slng.mintlify.site/voice-agents.md) is now a closed set: `bedrock-mantle/nvidia.nemotron-super-3-120b`, `bedrock-mantle/nvidia.nemotron-nano-3-30b`, and `groq/openai/gpt-oss-120b`. Pick one of these IDs when you create or update an agent so requests route to a live model.

## Moonshot Kimi K2 removed from Voice Agents

`groq/moonshotai/kimi-k2-instruct-0905` is no longer routable for Voice Agents. Switch existing agents and example payloads to `groq/openai/gpt-oss-120b` (or another model from the supported list) to keep calls working.

**May 28, 2026**

API v0.51.1

## Segment correlation for Cartesia Sonic 3

[Cartesia Sonic 3](https://slng.mintlify.site/api-reference/tts/cartesia-sonic-3/cartesia-sonic-3-ws.md) audio chunks now carry a `flush_id`. The counter starts at `0` and increments each time you send a `text` input with `flush: true` or a standalone flush message. Use it to map streamed audio frames back to the text segment that produced them.

## Pronunciation hints on Cognigy and Jambonz TTS bridges

The [Cognigy TTS bridge](https://slng.mintlify.site/api-reference/bridges/cognigy-tts-bridge/cognigy-tts-bridge-ws.md) and [Jambonz TTS bridge](https://slng.mintlify.site/api-reference/bridges/jambonz-tts-bridge/jambonz-tts-bridge-ws.md) accept an optional `pronunciation` object on `text` messages. Set `mode` (`ipa` or `ssml`), `name`, and `dictionary_id` to override how a term is spoken per chunk on providers that support [pronunciation dictionaries](https://slng.mintlify.site/pronunciation-dictionaries.md).

## Binary audio frames on the Unmute TTS bridge

The [Unmute TTS bridge](https://slng.mintlify.site/api-reference/unified-api/unmute-tts-bridge/unmute-tts-bridge-ws.md) now sends raw binary WebSocket frames instead of JSON `audio_chunk`, `segment_start`, and `segment_end` envelopes. The encoding and sample rate match the values you pass in the init message, so you can write the bytes straight to your audio pipeline. The `clear` control message has been removed; error and `audio_end` payloads now carry their fields under a `data` envelope (with `message` and `code` duplicated at the top level for backward compatibility).

**May 26, 2026**

API v0.50.1

## BYOK support for Kugel TTS

You can now bring your own Kugel API key when calling [Kugel 1 Turbo](https://slng.mintlify.site/api-reference/tts/kugel-1-turbo/kugel-1-turbo-ws.md), [Kugel 1](https://slng.mintlify.site/api-reference/tts/kugel-1/kugel-1-ws.md), and [Kugel 2](https://slng.mintlify.site/api-reference/tts/kugel-2/kugel-2-ws.md). Pass it on the `X-Slng-Provider-Key` header so Kugel bills your account directly, while the SLNG gateway's TTS cache still serves repeat requests. See [Bring your own key](https://slng.mintlify.site/execution-layer/byok.md) for the full provider list.

**May 23, 2026**

API v0.50.0

## HTTP streaming for Sarvam Bulbul v3

[Sarvam AI Bulbul Stream v3](https://slng.mintlify.site/api-reference/tts/sarvam-ai-bulbul-stream-v3/bulbul-stream-v3-http.md) is now available as a chunked HTTP endpoint. The response is raw audio bytes in the codec you set with `output_audio_codec` (there is no JSON envelope and no `X-Duration` header), so you can start playback as soon as the first chunk arrives. Use it to reach the same 30+ Indian-language voices as `bulbul:v3` with lower time-to-first-byte.

**May 22, 2026**

API v0.49.1

## Streaming transcription for Sarvam Saaras v3

Sarvam AI [Saaras v3](https://slng.mintlify.site/api-reference/stt/sarvam-ai-saaras/saaras-v3-ws.md) is now available over WebSocket for real-time transcription across 23 Indian languages. Configure each session with query parameters on the upgrade URL (`language-code`, `mode`, `sample_rate`, `input_audio_codec`, `high_vad_sensitivity`, and `vad_signals`) to tune voice activity detection and audio handling per stream.

**May 21, 2026**

API v0.48.1

## HTTP transcription for Nova 3 Indic languages

Deepgram Nova 3 now exposes HTTP endpoints for [Kannada](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-kannada-http), [Marathi](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-marathi-http), [Tamil](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-tamil-http), and [Telugu](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-telugu-http). You can transcribe these languages with a single request (binary upload or `url` field) instead of opening a WebSocket.

## Soniox TTS language coverage

[Soniox TTS](https://slng.mintlify.site/voices/soniox.md) now ships voices for 60+ languages, including Afrikaans, Bengali, Bulgarian, Catalan, Czech, Dutch, Greek, Gujarati, Hebrew, Indonesian, Kannada, Malay, Marathi, Norwegian, Tamil, Telugu, and more. Set the `language` field to any supported ISO 639-1 code on the [Soniox TTS HTTP endpoint](https://slng.mintlify.site/api-reference/tts/soniox-tts-v1/soniox-tts-v1-http.md) to pick a voice catalog.

## Runtime variables for voice agents

Voice agents can now declare `runtime_variables` so the model can capture values during a call and reuse them later in webhook URLs or system tool arguments. The built-in `set_runtime_variables` tool writes the values, which persist for the lifetime of the call or web session. See [Voice agents](https://slng.mintlify.site/voice-agents.md) for the setup pattern.

## New agent regions and world parts

The [region override](https://slng.mintlify.site/region-override) catalog now includes the `eu-non-eu` and `me` world parts, plus the `asia-south1`, `asia-southeast2`, and `australia-southeast1` regions. Pin requests to Sydney, Jakarta, or a non-EU European region with `X-Region-Override`, or stay inside the Middle East with `X-World-Part-Override: me`.

## Voice agents no longer need a separate API key

Voice agent create and duplicate requests no longer accept `slng_api_key`. Agents now use the API key you authenticate the request with. Drop the field from your payloads. Agent duplication also stops copying the inbound connection and call history; reconnect inbound routing on the copy if you need it.

## Deepgram Aura 2 region availability

SLNG-hosted [Deepgram Aura 2 English](https://slng.mintlify.site/api-reference/tts/deepgram-aura-2/aura-2-english-http.md) is no longer available in `us-east-1`. Route English TTS to `eu-north-1`. [Aura 2 Spanish](https://slng.mintlify.site/api-reference/tts/deepgram-aura-2/aura-2-spanish-http) drops the `na` world part and now serves only `eu`.

**May 20, 2026**

API v0.46.0

## Pronunciation dictionaries for TTS

You can now create reusable [pronunciation dictionaries](https://slng.mintlify.site/pronunciation-dictionaries.md) and attach them to any SLNG TTS request. Define rewrite rules once for brand names, acronyms, or domain terms, then reference the dictionary from HTTP, WebSocket, or Unified TTS calls. Manage dictionaries through the new [pronunciation dictionary endpoints](https://slng.mintlify.site/api-reference/tts/pronunciation-dictionaries/create-pronunciation-dictionary-http.md).

## Full voice catalogs on provider pages

Provider voice pages now list every voice in the catalog instead of capping the table at ten per language. The [Cartesia Sonic 3](https://slng.mintlify.site/voices/cartesia-sonic-3.md) page shows all 745 voices, [Sarvam Bulbul](https://slng.mintlify.site/voices/sarvam-bulbul.md) shows 405, [Deepgram Aura](https://slng.mintlify.site/voices/deepgram-aura.md) shows 91, [Soniox](https://slng.mintlify.site/voices/soniox.md) shows 120, [Murf](https://slng.mintlify.site/voices/murf.md) shows 111, and [Kugel](https://slng.mintlify.site/voices/kugel.md) shows 100. You can search and copy any voice ID directly from the provider page.

**May 19, 2026**

API v0.44.2

## Batch API reference now matches the gateway

The [Batch API reference](https://slng.mintlify.site/api-reference/speechmatics/create-batch-job.md) and [Batch API guide](https://slng.mintlify.site/batch-guide.md) are realigned with the live `api.batch.slng.ai` gateway. Request and response schemas, supported audio formats, and the three submission flows (direct upload, URL input, and presigned S3 upload) reflect what the service actually accepts.

**May 16, 2026**

API v0.44.1

## URL-based audio for HTTP transcription

The [Whisper Large v3](https://slng.mintlify.site/api-reference/stt/whisper-large-v3/whisper-large-v3-http.md), [Cognigy STT bridge](https://slng.mintlify.site/api-reference/bridges/cognigy-stt-bridge/cognigy-stt-bridge-http.md), [Jambonz STT bridge](https://slng.mintlify.site/api-reference/bridges/jambonz-stt-bridge/jambonz-stt-bridge-http.md), and [Unmute STT bridge](https://slng.mintlify.site/api-reference/unified-api/unmute-stt-bridge/unmute-stt-bridge-http.md) HTTP endpoints now accept a `url` field pointing to a publicly accessible audio file. Send a JSON body with `url` and `language` instead of a multipart upload when your audio is already hosted somewhere.

## Deepgram Nova 3 English region change

SLNG-hosted [Deepgram Nova 3 English](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-english-http.md) is no longer available in `eu-north-1`. Route English transcription to `australia-southeast1` or `us-east-1` instead.

**May 15, 2026**

API v0.43.3

## voiceai CLI

The new [`voiceai` CLI](https://slng.mintlify.site/sdks/cli) runs text-to-speech and speech-to-text from a terminal. Install it with `curl`, Homebrew, or `npm`, then pipe audio between SLNG models and other tools without writing an HTTP client.

## JavaScript and Python SDKs

The typed [JavaScript SDK](https://slng.mintlify.site/sdks/javascript.md) (`voiceai-sdk` on npm) supports Node, Bun, and Deno. The [Python SDK](https://slng.mintlify.site/sdks/python.md) (`voiceai-sdk` on PyPI) ships sync and async clients for Python 3.9+. Both wrap the full STT and TTS surface so you can drop the raw `fetch` and WebSocket plumbing.

## Agent skills for coding agents

The [`slng-ai/skills`](https://slng.mintlify.site/sdks/skills) pack teaches Claude Code and similar coding agents to call SLNG directly. Point your agent at the skills repo and it can pick models, build init messages, and stream audio for you.

## LiveKit Agents plugin

The [`livekit-plugins-slng`](https://slng.mintlify.site/agents/livekit-plugin) Python package connects [LiveKit Agents](https://docs.livekit.io/agents/) to any STT or TTS model on the SLNG gateway with a single configuration switch. You can swap providers or regions without changing your agent code.

## Embed a voice agent on the web

A new [browser embed guide](https://slng.mintlify.site/agents/embed-web.md) walks through adding a SLNG voice session to any web page. It uses LiveKit, a React frontend, and a backend proxy that keeps your API key off the client.

**May 14, 2026**

API v0.43.2

## Rime Arcana v3 Spanish

[Rime Arcana v3 (Spanish)](https://slng.mintlify.site/api-reference/tts/rime-arcana-v3/arcana-v3-spanish-ws) is available as a new TTS endpoint, with both streaming WebSocket and one-shot HTTP synthesis. Choose from ten Spanish voices (`aurelio`, `celestino`, `lark`, `luz`, `mar`, `nova`, `pola`, `seraphina`, `sirius`, and `ursa`) and pass `model: rime/arcana:3-es` on init.

## Rime Arcana v3 adds eu-north-1

[Arcana v3 English](https://slng.mintlify.site/api-reference/tts/rime-arcana-v3/arcana-v3-english-ws.md), [Hindi](https://slng.mintlify.site/api-reference/tts/rime-arcana-v3/arcana-v3-hindi-ws), and Spanish are now available in `eu-north-1`. You can route Arcana v3 synthesis to North Europe for lower latency in that region.

## Deepgram Nova 3 adds asia-south1

[Deepgram Nova 3](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-multi-language-ws) Tamil, Telugu, Marathi, and Kannada are now available in `asia-south1`, in addition to `ap-south-1`. You can route South Asian language transcription to the Mumbai GCP region.

## Deepgram Nova 3 Spanish region change

Deepgram Nova 3 Spanish is no longer available in `ap-southeast-2`. Use `australia-southeast1` or `us-east-1` instead.

## Soniox TTS region override removed

The `x-region` header is no longer accepted on [Soniox TTS v1](https://slng.mintlify.site/api-reference/tts/soniox-tts-v1/soniox-tts-v1-ws.md) requests. Soniox TTS runs only in `na`, so requests are routed there automatically.

**May 13, 2026**

API v0.42.2

## Expanded Murf Falcon voice catalog

The [Murf Falcon](https://slng.mintlify.site/voices/murf.md) voice catalog now lists 133 voices across more than 20 locales, including new entries for Bengali, Tamil, Telugu, Gujarati, Kannada, Punjabi, Japanese, Korean, Portuguese, Dutch, Polish, Greek, Croatian, and Scottish English. You can browse voices by full locale code and copy the `voice_id` for use in your init message.

## Simplified Unmute TTS bridge requests

The [Unmute TTS bridge](https://slng.mintlify.site/api-reference/unified-api/unmute-tts-bridge/unmute-tts-bridge-http.md) no longer requires a `model` field on HTTP or WebSocket init messages. The model is now inferred from the `{model_variant}` path. Send only `voice` and `text` for HTTP requests, or `voice` plus optional `config` on the WebSocket init.

**May 7, 2026**

API v0.42.1

## Rime Coda Indonesian TTS

[Rime Coda](https://slng.mintlify.site/api-reference/tts/rime-coda/coda-indonesian-ws.md) is available as a new TTS model in the `asia-southeast2` (Jakarta) region. It synthesizes Bahasa Indonesian with low latency across four voices (`pujianti_plesmita`, `siswoko_sigit`, `taryadi_dani`, and `usmany_tatianna`) and supports streaming WebSocket and one-shot HTTP synthesis.

## Region and world-part query parameters on bridges

The Cognigy, Jambonz, and Unmute HTTP bridges now accept `?region=` and `?world-part=` query parameters, mirroring the `X-Region-Override` and `X-World-Part-Override` headers. Use the query form when your platform cannot set custom headers; if both are present, the header wins. See [Integrations overview](https://slng.mintlify.site/integrations/overview.md#region-and-world-part-overrides) for details.

**May 6, 2026**

API v0.41.2

## Cartesia Sonic 3, Murf Falcon, Kugel, Soniox, Reson8, and Sarvam in the Unified API

The [Unified API](https://slng.mintlify.site/execution-layer/unified-api-models.md) now routes to Cartesia Sonic 3, Murf Falcon, KugelAudio Kugel 1/1-Turbo/2, Soniox TTS v1, Soniox Speech AI v4, Reson8 STT v1, and Sarvam Saaras v3. You can swap between these providers using a single request shape: pass identifiers like `cartesia/sonic:3`, `murf/murftts:falcon`, `kugelaudio/kugel:2`, `soniox/tts-rt:v1`, `soniox/speech-ai:rt-v4`, or `reson8/reson8stt:v1`.

## API Request tools support custom HTTP methods and raw payloads

[API Request tools](https://slng.mintlify.site/examples/agents-config.md#tools) accept `http_method` (`POST`, `PUT`, `PATCH`, or `DELETE`) and `webhook_format` (`envelope` or `raw`). Set `webhook_format` to `raw` to send only the tool arguments when the receiving service cannot parse the SLNG envelope.

## Tool execution tracking for voice agents

You can record webhook, template, human-transfer, and built-in tool activity against a call by posting to the [tool executions endpoint](https://slng.mintlify.site/api-reference/calls/submit-tool-execution). Each record carries the outcome, duration, and HTTP status, and submitted executions surface on the [call detail response](https://slng.mintlify.site/api-reference/calls/get-call.md) for debugging or analytics.

## ElevenLabs Flash v2.5 adds Asia Pacific region

[ElevenLabs Flash v2.5](https://slng.mintlify.site/api-reference/tts/eleven-flash-v2-5/elevenlabs-flash-v2-5-ws.md) is now available in `ap`, in addition to `eu`. You can route synthesis to Asia Pacific endpoints for lower latency in that region.

## Nova 3 multi-language adds eu-north-1

[Deepgram Nova 3 multi-language](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-multi-language-ws) is now available in `eu-north-1` as a specific region (in addition to the broader `eu` world part), giving you direct routing to North Europe.

## Deepgram Aura 2 English region change

[Deepgram Aura 2 English](https://slng.mintlify.site/api-reference/tts/deepgram-aura-2/aura-2-english-ws.md) is no longer available in `ap-southeast-2`. Use `eu-north-1` or `us-east-1` instead.

## Nova 3 Hindi region change

[Deepgram Nova 3 Hindi](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-hindi-ws.md) is no longer available in `ap-southeast-2`. Use `asia-south1` instead.

**April 30, 2026**

API v0.40.3

## Soniox TTS v1 general availability

[Soniox TTS](https://slng.mintlify.site/api-reference/tts/soniox-tts-v1/soniox-tts-v1-ws.md) graduates from preview to v1. Update your client to call `tts-rt:v1` and set `model` to `tts-rt-v1`. The `v1-preview` path and `tts-rt-v1-preview` model identifier are retired.

## Deepgram Aura 2 voice selection now required

The `model` field is now required on every [Deepgram Aura 2 English](https://slng.mintlify.site/api-reference/tts/deepgram-aura-2/aura-2-english-ws.md) and [Spanish](https://slng.mintlify.site/api-reference/tts/deepgram-aura-2/aura-2-spanish-ws) request. The previous defaults (`aura-2-thalia-en` and `aura-2-celeste-es`) no longer apply, so you must pick a voice explicitly.

**April 29, 2026**

API v0.40.1

## Kugel 2 TTS

[Kugel 2](https://slng.mintlify.site/api-reference/tts/kugel-2/kugel-2-ws.md) is available as a new TTS model in `eu`. It offers 87 voices with expressiveness control across 26 languages, including Arabic, Chinese, Hindi, Japanese, Korean, and Vietnamese.

## Soniox TTS v1-preview

[Soniox TTS v1-preview](https://slng.mintlify.site/api-reference/tts/soniox-tts-v1/soniox-tts-v1-ws.md) is available as a new TTS model in `na`, with both streaming WebSocket and one-shot HTTP synthesis. Browse the voice catalog on the [Soniox TTS voices](https://slng.mintlify.site/voices/soniox.md) page.

## Voice catalog pages for Cartesia Sonic 3 and Murf Falcon

You can now browse [Cartesia Sonic 3](https://slng.mintlify.site/voices/cartesia-sonic-3.md) and [Murf Falcon](https://slng.mintlify.site/voices/murf.md) voices with audio samples directly in the docs. Each entry shows the `voice_id` to pass in your init message.

## Nova 3 multi-language adds EU region

[Deepgram Nova 3 multi-language](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-multi-language-ws) is now available in `eu`, in addition to `ap-southeast-2` and `us-east-1`. You can route multilingual transcription to European endpoints for lower latency.

## Nova 3 Hindi region change

[Deepgram Nova 3 Hindi](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-hindi-ws.md) is no longer available in `ap-south-1`. Use `ap-southeast-2` or `asia-south1` instead.

**April 25, 2026**

API v0.38.0

## Soniox Speech AI Real-time v4

Soniox Speech AI moves to v4. Use the [Speech AI Real-time v4](https://slng.mintlify.site/api-reference/stt/soniox-speech-ai-real-time-v5/speech-ai-real-time-v5-ws.md) endpoint for streaming transcription with speaker diarization, automatic language identification, and configurable endpoint detection across 60+ languages. The v3 endpoint has been retired. Point clients at the new path to continue receiving native Soniox token frames.

## LiveKit plugin compatibility refresh

The [LiveKit Agents plugin](https://slng.mintlify.site/agents/livekit-plugin.md) now targets `livekit-agents>=1.5.1` and Python 3.10+. You can pass model-specific options as keyword arguments. For example, `whisper_params` for Whisper, `target_language_code` for Sarvam STT, or `modelId` and `speakingStyle` for Rime Arcana. New `slng_base_url` and `http_session` arguments let you point at a self-hosted gateway and reuse an `aiohttp.ClientSession`.

## Sarvam Saaras v3 STT not supported in LiveKit plugin

Saaras is HTTP-only on SLNG and has no WebSocket endpoint, so it cannot run through the LiveKit plugin's realtime path. For Hindi voice agents, use `slng/deepgram/nova:3-hi` or `slng/deepgram/nova:3-multi`. See the [LiveKit plugin provider notes](https://slng.mintlify.site/agents/livekit-plugin.md#provider-notes) for details.

**April 24, 2026**

API v0.37.0

## Expanded regions for Murf Falcon TTS

[Murf Falcon](https://slng.mintlify.site/api-reference/tts/murf-falcon/murf-falcon-ws.md) is now available in `ap`, `eu-non-eu`, `me`, and `na`, in addition to the existing `eu` world part. You can now route synthesis closer to users across the Americas, Asia Pacific, and the Middle East.

## Asia Pacific region for Soniox Speech AI Real-time v3

[Soniox Speech AI Real-time v3](https://slng.mintlify.site/api-reference/stt/soniox-speech-ai-real-time-v5/speech-ai-real-time-v5-ws.md) adds the `ap` world part alongside `eu` and `na`. Route transcription to Asia Pacific endpoints for lower latency in that region.

**April 23, 2026**

API v0.36.0

## URL and presigned S3 inputs for Batch STT

You can now submit audio to the [Batch STT API](https://slng.mintlify.site/batch-guide.md) without uploading a file on every request. Pass a publicly accessible `input_url`, or request a presigned S3 URL, upload directly, then create the job with the returned `s3_key`. Both methods accept an optional `metadata` object for attaching arbitrary key-value pairs to a job.

## Batch API usage guide

A new [Batch API guide](https://slng.mintlify.site/batch-guide.md) walks through the three input methods (file upload, URL input, and presigned S3 upload) with request flows and sample payloads.

## Deepgram Aura 2 English in eu-north-1

[Deepgram Aura 2 English TTS](https://slng.mintlify.site/api-reference/tts/deepgram-aura-2/aura-2-english-ws.md) is now available in `eu-north-1`, in addition to `ap-southeast-2` and `us-east-1`.

## Whisper Large v3 Compressed removed

The Whisper Large v3 Compressed STT model has been retired from the catalog. Use [Whisper Large v3](https://slng.mintlify.site/api-reference/stt/whisper-large-v3/whisper-large-v3-ws.md) for multilingual transcription going forward.

**April 22, 2026**

API v0.32.0

## Runtime variables for voice agents

Voice agents can now capture values during a call and reuse them in API Request URLs and automatic tool arguments. Define a `runtime_variables` array on your agent, and the model sets values through the built-in `set_runtime_variables` tool. See the [agent configuration examples](https://slng.mintlify.site/examples/agents-config.md#runtime-variables) for setup details.

## API Request HTTP method and payload format

API Request tools on voice agents accept `http_method` (POST, PUT, PATCH, or DELETE) and `webhook_format` (`envelope` or `raw`). Use `raw` to send only the tool arguments as the request body, skipping the SLNG metadata envelope. See the [shared tool guide](https://slng.mintlify.site/examples/agents-config.md#tools).

## Expanded regions for Rime Arcana v2 and Cartesia Sonic 3

[Rime Arcana v2 TTS](https://slng.mintlify.site/api-reference/tts/rime-arcana-v2/arcana-v2-english-ws.md) is now available in `eu-north-1` and `us-east-1`, in addition to `ap-southeast-2`. [Cartesia Sonic 3 TTS](https://slng.mintlify.site/api-reference/tts/cartesia-sonic-3/cartesia-sonic-3-ws.md) is now available in all three world parts: `ap`, `eu`, and `na`.

## New regions for Deepgram Nova 3 English and Hindi

[Nova 3 English](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-english-ws.md) is now available in `ap-south-1` and `us-east-1`. [Nova 3 Hindi](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-hindi-ws.md) adds `asia-south1` alongside existing regions.

## Utterance end events on Unmute STT Bridge

The [Unmute STT Bridge](https://slng.mintlify.site/api-reference/unified-api/unmute-stt-bridge/unmute-stt-bridge-ws.md) now emits `utterance_end` events when the upstream model signals the end of a spoken utterance. This gives you an explicit boundary marker for segmenting transcription output.

## Native token stream for Soniox Speech AI

[Soniox Speech AI Real-time v3](https://slng.mintlify.site/api-reference/stt/soniox-speech-ai-real-time-v5/speech-ai-real-time-v5-ws.md) now returns native Soniox token frames instead of normalized transcripts. You receive interim and final tokens directly, including `<end>` and `<fin>` endpoint markers when endpoint detection is enabled.

**April 20, 2026**

API v0.31.3

## Attachment values for voice agents

Tool attachments can use `{{variable}}` value references in supported fields, including API Request URLs, automatic tool arguments, transfer destinations, and timezones. Values resolve at dispatch or when the tool runs. Current value sources and examples are documented on the [Configuration](https://slng.mintlify.site/examples/agents-config.md#attachment-values) page.

## Tool execution tracking on agent calls

A new [tool executions endpoint](https://slng.mintlify.site/api-reference/calls/submit-tool-execution) lets you record webhook, template, human transfer, and built-in tool activity against a call. Execution records (including outcome, duration, and HTTP status) also appear in the [call detail response](https://slng.mintlify.site/api-reference/calls/get-call.md).

**April 18, 2026**

API v0.31.2

## Cartesia Sonic 3 TTS

[Cartesia Sonic 3](https://slng.mintlify.site/api-reference/tts/cartesia-sonic-3/cartesia-sonic-3-ws.md) is available as a new TTS provider. It supports low-latency streaming synthesis over WebSocket with context-aware generation controls.

## Reson8 STT

[Reson8 STT v1](https://slng.mintlify.site/api-reference/stt/reson8-stt-v1/reson8-stt-v1-ws.md) is available as a new STT provider. It supports real-time transcription over WebSocket with word-level timestamps, confidence scores, and partial results in nine languages including Dutch, French, German, and Spanish.

## Deepgram Nova 3 Indic language endpoints

Four new SLNG-hosted Deepgram Nova 3 language variants are available in `ap-south-1` (Mumbai): [Kannada](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-kannada-ws.md), [Marathi](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-marathi-ws.md), [Tamil](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-tamil-ws), and [Telugu](https://slng.mintlify.site/api-reference/stt/deepgram-nova-3/nova-3-telugu-ws.md). Each has a dedicated WebSocket endpoint for that language.

## Soniox Speech AI version correction

The Soniox STT endpoint is now correctly labeled [Speech AI Real-time v3](https://slng.mintlify.site/api-reference/stt/soniox-speech-ai-real-time-v5/speech-ai-real-time-v5-ws.md). URLs and navigation have been updated accordingly.

**April 15, 2026**

API v0.22.1

## Batch speech-to-text API

You can now transcribe audio files asynchronously with the new [Batch STT API](https://slng.mintlify.site/api-reference/batch.md). Upload a file, poll the job status, and download the transcript when ready. Supported formats include wav, mp3, flac, aac, ogg, m4a, mp4, amr, and mpeg. Powered by Speechmatics.

**April 11, 2026**

API v0.22.0

## Murf Falcon TTS

[Murf Falcon](https://slng.mintlify.site/api-reference/tts/murf-falcon/murf-falcon-ws.md) is available as a new TTS provider. It supports multilingual speech synthesis over WebSocket with multiple encodings and sample rates.

## Unified API documentation

The new [Unified API](https://slng.mintlify.site/execution-layer/unified-api) section explains how to use one endpoint pattern for every STT and TTS model. Swap providers by changing only the URL path. Your auth, request format, and code stay the same. Includes guides on [parameter coverage](https://slng.mintlify.site/execution-layer/unified-api-parameters.md) and [supported models](https://slng.mintlify.site/execution-layer/unified-api-models.md).

## Integrations hub

A new [Integrations](https://slng.mintlify.site/integrations/overview.md) page lists third-party platforms you can connect to SLNG. LiveKit, Cognigy, and Jambonz each have dedicated setup paths.

## Whisper Large v3 endpoint consolidated

The separate Whisper Large v3 Compressed endpoint has been removed. Use the standard [Whisper Large v3](https://slng.mintlify.site/api-reference/stt/whisper-large-v3/whisper-large-v3-ws.md) endpoint, which now handles compressed audio directly.

**April 9, 2026**

API v0.20.0

## Language selection for Nova 3 STT

SLNG-hosted Deepgram Nova 3 STT endpoints accept a `language` parameter in the WebSocket `init` config. Supported locales by variant:

* **English**: `en`, `en-au`, `en-us`, `en-nz`, `en-gb`, `en-in`
* **Spanish**: `es`, `es-us`, `es-419`, `es-ar`, `es-mx`, `es-es`
* **Hindi**: `hi`, `en`
* **Multi-language**: `multi`

The Hindi variant also accepts `en`, so you can transcribe English audio without switching endpoints. See the [Speech-to-Text models page](https://slng.mintlify.site/models/stt.md) for the full parameter list.

## More sample rates for Rime Arcana TTS

[Rime Arcana](https://slng.mintlify.site/voices/rime-arcana.md) now supports 8, 16, 22.05, 24 (default), 32, 44.1, and 48 kHz. You can match your audio pipeline directly without resampling.

## Simplified endpointing parameter

The `endpointing` parameter on Deepgram STT endpoints now accepts only an integer (milliseconds of silence before finalizing speech). Set it to `0` to disable. Default remains `10`.

**April 8, 2026**

API v0.19.0

## Graceful WebSocket session close

Send `{ "type": "close" }` on any [WebSocket](https://slng.mintlify.site/websockets.md) connection to shut down cleanly. The server finishes processing remaining audio, then closes. This replaces the previous `cancel` behavior and works across TTS, STT, and [bridges](https://slng.mintlify.site/execution-layer/unified-api).

## Keepalive for STT streams

Send `{ "type": "keepalive" }` on STT WebSocket connections to prevent idle timeouts during pauses. Useful for voice agent sessions where the user goes silent but the connection should stay open.

## Endpointing controls for Deepgram Nova STT

Two new parameters on [Deepgram Nova STT models](https://slng.mintlify.site/models/stt.md) for tuning speech segmentation:

* **`endpointing`**: milliseconds of silence before finalizing speech. Set to `false` to disable. Default: `10`.
* **`utterance_end_ms`**: milliseconds of silence between words before an `UtteranceEnd` event. Range: 200–5000 ms, default: 1000 ms.

## India region for Nova 3 Hindi

Deepgram Nova 3 Hindi is now available in `ap-south-1` (Mumbai), alongside `ap-southeast-2` (Sydney). Use the `X-Region-Override` header to route to the closest region. See [models by region](https://slng.mintlify.site/models/by-region.md).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.