About AssemblyAI
AssemblyAI offers speech AI models that you call through an API. The models handle transcription, real-time speech recognition and text-to-speech for voice applications.
What it does
AssemblyAI turns recorded or live audio into text, and it can also pull insights out of that text. A separate set of models speaks text aloud for voice agents and automated phone menus. Developers send audio to the API and get back a transcript or a generated audio file.
Key features
- Transcription of recorded audio in 32 languages
- Real-time speech-to-text for live voice agent calls
- Text-to-speech voices for agents and phone menus
- Support for speakers who switch languages mid-conversation
- Speaker labels for recordings with several people
Who it's for
Developers building voice agents, call analytics, meeting notes, captioning or media search tools. Teams that want speech models inside their own product will find the API the main way in.
Pricing and good to know
Pricing is on the company's pricing page. Text-to-speech covers six languages, fewer than the speech-to-text range, so check your target language before you build on it. The models are offered as a hosted API that your code calls.
Highlights
- Transcribe recorded audio across 32 languages
- Real-time speech-to-text for voice agent calls
- Text-to-speech voices for agents and phone menus
- Handles speakers who switch languages mid-conversation
- Speaker labels for recordings with several people
