AssemblyAI
Not clearly publishedVoice AI infrastructure for builders of speech-to-text and voice applications
工具信息
什么是 AssemblyAI？
AssemblyAI is a Voice AI infrastructure platform that provides production-ready Speech AI models, APIs, and infrastructure. It enables transcription of pre-recorded, realtime, and sync audio, extraction of structured insights from speech, and powering of voice agents and related applications through speech-to-text and audio intelligence capabilities.
核心功能
Pre-recorded Speech-to-Text API
Turns raw audio into structured, actionable intelligence using purpose-built models in a single API call. Supports Universal-3.5 Pro and Universal-2, speaker diarization, natural language prompting, key terms, 99+ languages, PII reduction, LLM-powered insights, and a Medical Mode add-on for healthcare terminology with HIPAA and BAA available.
Realtime Speech-to-Text API
Transcribes live audio and video in real time with ultra-low latency and high accuracy. Supports U3.5 Pro Realtime, Universal Streaming, Whisper Streaming, approximately 150ms latency, end-of-turn detection, key terms prompting, 18+ languages, and speaker diarization.
Sync Speech-to-Text API
Delivers finished transcripts in a single API call for short clips with no polling required; send audio and read the transcript from the response.
Voice Agent API
End-to-end voice agent infrastructure built on the Realtime Speech-to-Text API. Features entity accuracy, turn detection, key terms, 18+ languages, short-utterance handling, and Medical Mode add-on.
Speech Understanding API
Converts transcripts into structured intelligence as part of the understand product suite.
LLM Gateway
Component for connecting speech outputs to large language models within the understand suite.
Guardrails
Safety and guardrails capabilities for voice AI workflows.
AssemblyAI Cloud
Cloud-based Voice AI infrastructure that scales with usage.
Self-Hosted Voice AI
Self-hosted deployment option for Voice AI models and infrastructure.
Universal-3.5 Pro
Flagship speech-to-text model that handles real-world audio for both real-time and pre-recorded use, available as the latest release.
使用场景
- #1
Voice Agents
Build end-to-end voice agent applications using realtime speech-to-text and supporting infrastructure.
- #2
AI Notetakers
Power AI-powered notetaking products from spoken audio.
- #3
AI Scribe
Support AI scribe applications, including medical contexts.
- #4
Conversation Intelligence
Extract insights and structured data from conversations.
- #5
Agent Assist
Enable real-time assistance for agents in conversations.
- #6
Call Analytics
Analyze calls for insights using speech understanding capabilities.
- #7
Medical Transcription
Optimize transcription for medical terminology and healthcare conversations with improved accuracy; HIPAA and BAA available.
- #8
Dictation
Support dictation workflows from audio input.
定价方案
Try the API for free. Marketing materials state pricing that does not turn against users at scale. No specific plans or rates are detailed on the source pages.
