What it does
AssemblyAI provides speech AI as developer APIs. Its speech-to-text APIs cover pre-recorded audio, real-time streaming and short synchronous clips, with transcripts in 99 languages. A Speech Understanding API goes beyond transcription to extract speaker identity, sentiment, chapters and summaries in a single call. Other products include a Voice Agent API with built-in turn detection and interruption handling, a Dictation API that returns text cleaned of filler words and self-corrections, and Guardrails that redact personal data and moderate content in audio and transcripts. An LLM Gateway routes requests between several LLMs from one endpoint with fallback. A Text-to-Speech API is listed as coming soon, and self-hosted and cloud deployments are offered. The site lists use cases such as voice agents, AI notetakers, call analytics, medical transcription and captioning. Developers get documentation, an API reference, cookbooks, a changelog and a no-code playground for trying the models.
Capabilities
- Pre-recorded, real-time and synchronous speech-to-text APIs
- Speech Understanding for speaker identity, sentiment, chapters and summaries
- Voice Agent API with turn detection and interruption handling
- Guardrails for redacting personal data and moderating content
- LLM Gateway for routing between LLMs with fallback
- Self-hosted and cloud deployment options
When to use it
- Adding transcription to a product through an API
- Building a voice agent or an AI notetaker
- Analysing recorded calls for speakers, sentiment and summaries
- Producing captions or medical dictation from audio
