llms.txt
# SpeechSDK
> The unified text-to-speech SDK for JavaScript and TypeScript. One open-source interface across OpenAI, ElevenLabs, Deepgram, Cartesia, Google, Hume, Mistral, and 10+ more providers — bring your own API keys.
SpeechSDK (`@speech-sdk/core`) is a free, open-source toolkit for building AI audio applications without locking into a single voice provider. Use the same `generateSpeech`, `streamSpeech`, and `generateConversation` calls across every supported provider.
## Features
- **One API, every provider** — unified model strings, consistent response format, BYO API keys. Includes streaming, auto-retries, error handling, and provider-specific options.
- **Multi-speaker conversations** — generate a multi-voice conversation in one call, mixing voices across providers, with automatic volume leveling and turn stitching.
- **Auto-chunking & timestamps** — sentence-boundary splitting for inputs that exceed a provider's per-request limit, word-level timestamps, and SRT/VTT caption export.
- **Voice cloning, audio tags, pronunciations, speed control** — supported on every provider that exposes them.
## Related
- [Jellypod](https://jellypod.com), the team that builds and maintains SpeechSDK.
## Documentation
- **SpeechSDK**
- [Getting Started](/docs): Get started with SpeechSDK — the universal text-to-speech SDK for JavaScript and TypeScript.
- Providers
- [All Providers](/docs/providers): All supported text-to-speech providers, their models, and configuration.
- [OpenAI](/docs/providers/openai): OpenAI text-to-speech models (gpt-4o-mini-tts, tts-1, tts-1-hd).
- [ElevenLabs](/docs/providers/elevenlabs): ElevenLabs TTS models (eleven_v3, multilingual v2, flash variants).
- [Deepgram](/docs/providers/deepgram): Deepgram Aura text-to-speech.
- [Cartesia](/docs/providers/cartesia): Cartesia Sonic text-to-speech with SSML, voice cloning, and audio tags.
- [Hume](/docs/providers/hume): Hume Octave text-to-speech with expressive voices and voice cloning.
- [Google (Gemini TTS)](/docs/providers/google): Google Gemini 3.1 Flash and 2.5 Flash/Pro text-to-speech preview models.
- [Fish Audio](/docs/providers/fish-audio): Open-source Fish Audio S2 with audio tags and voice cloning.
- [Inworld](/docs/providers/inworld): Inworld text-to-speech with streaming and multilingual voices.
- [Murf](/docs/providers/murf): Murf GEN2 and FALCON text-to-speech.
- [Resemble](/docs/providers/resemble): Open-source Resemble AI text-to-speech with voice cloning.
- [Smallest AI](/docs/providers/smallest-ai): Smallest AI Lightning text-to-speech models with multilingual voices and configurable output formats.
- [fal](/docs/providers/fal): fal.ai TTS model marketplace with voice cloning on select models.
- [Mistral (Voxtral TTS)](/docs/providers/mistral): Mistral Voxtral open-source text-to-speech with voice cloning.
- [xAI (Grok)](/docs/providers/xai): xAI Grok text-to-speech with native audio tags and BCP-47 language control.
- **Features**
- [Streaming](/docs/streaming): Stream text-to-speech audio chunk-by-chunk with streamSpeech for low-latency playback.
- [Multi-Speaker Conversation](/docs/multi-speaker-conversation): Generate a single audio file from a multi-turn, multi-voice script with generateConversation.
- [Auto-Chunking](/docs/auto-chunking): Generate speech from inputs that exceed a provider's per-request character limit by splitting on sentence boundaries and stitching the audio.
- [Timestamps & Captions](/docs/timestamps): Return word-level timestamps from generateSpeech and convert them into SRT or WebVTT captions.
- [Audio Tags](/docs/audio-tags): Write expressive audio cues once and let SpeechSDK translate them for each provider.
- [Output Formats](/docs/output-formats): Request wav, mp3, or pcm audio from generateSpeech and generateConversation.
- [Pronunciations](/docs/pronunciations): Customize how specific words are pronounced via text substitution rules and gateway-managed dictionaries, with timestamps mapped back to your original text.
- [Speed Control](/docs/speed): Slow down or speed up generated audio with the `speed` parameter on generateSpeech and generateConversation.
- [Voice Cloning](/docs/voice-cloning): Clone voices using reference audio with supported providers.
- **Usage**
- [Configuration](/docs/configuration): Custom API keys, base URLs, and fetch implementations.
- [Speech Result](/docs/result): Working with the audio data returned by generateSpeech.
- [Error Handling](/docs/error-handling): Handle API errors, empty responses, and other failure modes.
