Changelog

This entry adds Sarvam AI as a text-to-speech and speech-to-text vendor, for Indian English, Hindi and other Indic languages.

New Features & Improvements

1. Sarvam AI text-to-speech — sarvam is available as a TTS vendor, with streaming TTS and speech credentials in the portal and API. The Bulbul models bulbul:v4-flash (default), bulbul:v3 and bulbul:v2 cover eleven Indian languages, and the portal lists the bulbul:v4-flash personas by language. Options are pace, pitch, loudness, temperature, enable_preprocessing and dict_id, plus buffering controls for streaming. See Sarvam AI.
2. Sarvam AI speech-to-text — the same sarvam credential serves realtime speech recognition with saaras:v3-realtime (default) or saaras:v4, in 23 Indian languages or with automatic language detection. Sarvam signals the start and end of speech and returns a final transcript after each pause. sarvamOptions sets the stream type, the transcript mode (including translation to English), the voice activity settings, a prompt and, for saaras:v4, key terms. See recognizer.

This entry adds Smallest AI as a text-to-speech and speech-to-text vendor, for English, Indic, European and Asian languages.

New Features & Improvements

1. Smallest AI text-to-speech — smallestai is available as a TTS vendor, with streaming TTS and speech credentials in the portal and API. The Lightning models lightning_v3.1_pro (default), lightning_v3.1_pro_10_26 and lightning_v3.1 are supported, and the portal lists the Pro voices by language. Options are speed, language, number_pronunciation_language, date_format, math_notation, pronunciation_dicts and content_filter, plus buffering controls for streaming, and word-level timestamps make alignment events available for the voices that return them. See Smallest AI.
2. Smallest AI speech-to-text — the same smallestai credential serves realtime speech recognition with Pulse (pulse, default, or the English-only pulse-2). hints are sent as Pulse’s keywords (up to 100 terms), and smallestaiOptions sets formatting, end-of-utterance timing, diarization, redaction, timestamps and detection features. See recognizer.

This entry adds Azure OpenAI as a text-to-speech and speech-to-text vendor, using the OpenAI speech models deployed on your own Azure OpenAI resource.

New Features & Improvements

1. Azure OpenAI text-to-speech — azureopenai is available as a TTS vendor for one-shot say (no TTS streaming), with speech credentials in the portal and API. A credential holds the resource key, the resource endpoint and the deployment names. The default deployment is gpt-4o-mini-tts with the voice alloy, and options are speed, instructions and model_id. See Azure OpenAI.
2. Azure OpenAI speech-to-text — the same azureopenai credential serves realtime speech recognition on a transcription deployment (gpt-live-transcribe by default, or gpt-realtime-whisper, gpt-4o-transcribe, gpt-4o-mini-transcribe). azureopenaiOptions mirrors openaiOptions: deployment, model family, prompt, languages, keywords, delay, noise reduction, turn detection and local voice activity settings. See recognizer.

This entry adds Gnani as a text-to-speech and speech-to-text vendor, for Indian English, Hindi, Hinglish and other Indic languages.

New Features & Improvements

1. Gnani text-to-speech — gnani is available as a TTS vendor for one-shot say (no TTS streaming), with speech credentials in the portal and API. The Timbre models timbre-v2.5 (default) and timbre-v3.0 cover ten Indian languages plus Hinglish, and the portal lists the voices by language. Options are model_id, language and speed. See Gnani.
2. Gnani speech-to-text — the same gnani credential serves realtime speech recognition with prisma-v2.5 (default) or prisma-v3.0, in ten Indian languages or with automatic Indic detection. Gnani signals the start and end of speech and returns one final transcript per utterance. hints are sent as bias words or key terms, and gnaniOptions sets the transcript format, substitutions, context, idle timeout and voice activity settings. See recognizer.

This entry adds Meta as a speech-to-text vendor, in 25 languages with code-switching.

New Features & Improvements

1. Meta speech-to-text — meta is available as an STT vendor, with speech credentials in the portal and API, for realtime speech recognition with Meta’s muse-voice-transcribe-1.0. Meta signals the start of speech, sends interim transcripts and returns one final transcript per turn. The recognizer’s language is sent as a language bias, hints are sent as keywords, and metaOptions sets the mode, the partial transcript style, keywords, language bias and a session id. See recognizer.

This entry adds MiniMax as a text-to-speech vendor, with voices in more than twenty languages.

New Features & Improvements

1. MiniMax text-to-speech — minimax is available as a TTS vendor, with streaming TTS and speech credentials in the portal and API. The Speech models speech-2.8-turbo (default) through speech-01-hd are available, and the portal lists MiniMax’s system voices by language. Options include speed, volume, pitch, emotion, text normalization, language boost, voice effects, pronunciation dictionaries and voice mixing. Streaming sends text over a live connection and cancels on barge-in. See MiniMax.

This entry adds Neuphonic as a text-to-speech vendor, with voices in more than forty languages.

New Features & Improvements

1. Neuphonic text-to-speech — neuphonic is available as a TTS vendor, with streaming TTS and speech credentials in the portal and API. The portal lists Neuphonic’s stock voices by language, and account voices can be used by their voice id. Options are language, speed, temperature, ipa_key and, for streaming, timestamps. Streaming returns word-level timestamps, so alignment events are available. See Neuphonic.

This entry adds Camb.ai as a text-to-speech vendor, with MARS voices in more than thirty locales.

New Features & Improvements

1. Camb.ai text-to-speech — camb is available as a TTS vendor, with streaming TTS and speech credentials in the portal and API. Models are mars-flash (default), mars-pro, mars-instruct and the mars-8.1 betas; voices are numeric ids, and cloned voices work by their id. Options include speaking_rate, enhancement switches, inference_steps and the mars-instruct user_instructions. Streaming returns word-level timestamps, so alignment events are available. See Camb.ai.

This entry adds Async as a text-to-speech vendor, with voices in up to fifteen languages.

New Features & Improvements

1. Async text-to-speech — asyncai is available as a TTS vendor, with streaming TTS and speech credentials in the portal and API. Models are async_flash_v1.5 (default), async_flash_v1.0 and async_pro_v1.0; voices are Async voice ids. Options are model_id, language, speed_control and stability. Streaming does not return word timestamps, so alignment events are not available. See Async.

This entry adds Respeecher as a text-to-speech vendor, with real-time models in six languages.

New Features & Improvements

1. Respeecher text-to-speech — respeecher is available as a TTS vendor, with streaming TTS and speech credentials in the portal and API. Models are en-rt (default), ua-rt, ar-rt, de-rt, pt-br-rt and sv-rt, picked from the language unless model_id is set. Options are model_id and the sampling controls seed, temperature, top_k, top_p, min_p, presence_penalty, repetition_penalty and frequency_penalty. Streaming does not return word timestamps, so alignment events are not available. See Respeecher.

This entry adds Bland AI as a text-to-speech vendor.

New Features & Improvements

1. Bland AI text-to-speech — bland is available as a TTS vendor, with streaming TTS and speech credentials in the portal and API. The voice is a Bland voice UUID, which also selects the model; Karen is the default en-US voice. Options are expressiveness and stability. Streaming does not return word timestamps, so alignment events are not available. See Bland AI.

This entry adds Simplismart as a text-to-speech vendor.

New Features & Improvements

1. Simplismart text-to-speech — simplismart is available as a TTS vendor for one-shot say (no TTS streaming), with speech credentials in the portal and API. Models are Orpheus 3B (default), Veena for Hindi and Qwen 3 TTS on a dedicated deployment, set with the credential’s API host. Options are model_id, temperature, top_p, repetition_penalty and max_tokens, plus language and leading_silence for Qwen. See Simplismart.

This entry adds Spitch as a text-to-speech vendor.

New Features & Improvements

1. Spitch text-to-speech — spitch is available as a TTS vendor for one-shot say (no TTS streaming), with speech credentials in the portal and API. Voices cover English, Yoruba, Hausa, Igbo, Amharic and Nigerian Pidgin; lina is the default en-US voice. Options are speed and api_uri. See Spitch.

This entry adds Kokoro as a text-to-speech vendor, for self-hosted Kokoro servers.

New Features & Improvements

1. Kokoro text-to-speech — kokoro is available as a TTS vendor for one-shot say (no TTS streaming), with speech credentials in the portal and API. jambonz calls your server’s OpenAI-compatible /v1/audio/speech endpoint; the credential needs the server’s API base URL, and the API key is optional. The portal lists the bundled voices for nine languages; af_heart is the default en-US voice. Options are model_id, lang_code, speed, volume_multiplier, normalization_options, voice_aliases, allow_voice_tags, ssml, api_uri and api_key. See Kokoro.

This entry adds Vakyam as a text-to-speech vendor, for Indian English and Indic languages.

New Features & Improvements

1. Vakyam text-to-speech — vakyam is available as a TTS vendor for one-shot say (no TTS streaming), with speech credentials in the portal and API. The raaga-v1 model covers en-IN, hi-IN, ta-IN, te-IN, kn-IN, mr-IN, gu-IN and bn-IN, and the portal lists 16 voices for every language; Archana is the default. Other jambonz languages map to the matching xx-IN code, falling back to en-IN. Options are model_id, speed, language and api_uri. See Vakyam.

This entry adds Groq as a text-to-speech vendor, using Canopy Labs’ Orpheus models.

New Features & Improvements

1. Groq text-to-speech — groq is available as a TTS vendor for one-shot say (no TTS streaming), with speech credentials in the portal and API. canopylabs/orpheus-v1-english covers en-US (default voice autumn) and canopylabs/orpheus-arabic-saudi covers ar-SA (default voice fahad); the model is picked by language unless model_id is set. Options are model_id, speed and sample_rate; sentences longer than 200 characters are split. A Groq speech credential is separate from a Groq LLM service. See Groq.

This entry adds Hume Octave as a text-to-speech vendor.

New Features & Improvements

1. Hume text-to-speech — hume is available as a TTS vendor, with streaming TTS and speech credentials in the portal and API. The default voice is Ava Song; a voice UUID or a voice from your own account (voice_provider: CUSTOM_VOICE) can be entered as a custom voice. Options are model_id (Octave 1 or 2), voice_provider, description, speed, trailing_silence, instant_mode, temperature, context_generation_id and split_utterances. Hume has no language setting, and alignment events are not available. Hume no longer lists the Octave TTS docs, so confirm availability with Hume before production use. See Hume.

This entry adds speech-to-text to the existing Inworld vendor, in 30 languages.

New Features & Improvements

1. Inworld speech-to-text — the inworld credential, until now TTS only, can be used for realtime speech recognition with inworld/inworld-stt-1. A credential used only for STT no longer needs a TTS model. Inworld signals the start of speech, sends interim transcripts and returns one final transcript per turn. The recognizer’s language is sent as its primary subtag, or omitted for automatic detection; hints are sent as prompts (up to 100), and inworldOptions sets word timestamps, diarization, end-of-turn confidence and silence, voice activity threshold, inactivity timeout and voice profile. See recognizer and Inworld.

This entry adds text-to-speech to the existing Speechmatics vendor, for British and American English.

New Features & Improvements

1. Speechmatics text-to-speech — the speechmatics credential, until now STT only, can be used for one-shot text-to-speech say (no TTS streaming). A credential used only for TTS no longer needs the STT endpoint. The preview voices are sarah (default) and theo for en-GB, and megan and jack for en-US; synthesizer.language is not sent. Options are sm_app and api_uri. Speechmatics TTS is a preview product. See Speechmatics.

This entry adds text-to-speech to the existing Soniox vendor, in sixty languages.

New Features & Improvements

1. Soniox text-to-speech — the soniox credential, until now STT only, can be used for text-to-speech, with streaming TTS. The credential gains an optional TTS model (tts-rt-v2, the default) and an optional TTS host for regional endpoints. The portal lists 28 voices, each speaking every language; Adrian is the default, and synthesizer.language is sent as its bare language code. Options are model_id, language, speed, reduce_silence and client_reference_id. Streaming sends text over a live connection, cancels on barge-in and returns character-level timestamps, so alignment events are available. See Soniox.

TTS streaming is now limited to vendors whose websocket accepts text token by token.

Changes

1. Speechify, Giggy and Mistral no longer stream TTS — these vendors have no token-streaming websocket, so say with stream: true and the agent verb no longer use them; one-shot say is unchanged. The new HTTP-only vendors above (Azure OpenAI, Gnani, Simplismart, Spitch, Kokoro, Vakyam, Groq and Speechmatics) are one-shot say only for the same reason.