assemblyai-streaming

Speech-to-text streaming connects to AssemblyAI's WebSocket endpoint at wss://streaming.assemblyai.com/v3/ws using your API key. Stream audio in real time to receive low-latency transcripts as words are spoken, ideal for live meetings, voice agents, and microphone input without waiting for recording completion.

Yes, AssemblyAI's REST API at https://api.assemblyai.com/v2/transcript handles async transcription for pre-recorded files. Submit audio, poll for completion, and retrieve full transcripts—useful when immediate processing isn't required and batch workflows are preferred.

WebSocket streaming delivers transcript events in real time as audio arrives, meeting ultra-low-latency needs for live interaction. REST async transcription processes uploaded or batched audio without live streaming, trading immediacy for simpler integration and fire-and-forget workflows.

Use AssemblyAI's LLM Gateway at https://llm-gateway.assemblyai.com to route transcripts to Claude, GPT, or Gemini for Q&A, summarization, or analysis. Combine streaming STT output with LLM processing for voice agents and meeting-note automation in one workflow.

The Skill runs in Python and Bash environments with API-key authentication. Python and shell scripts can manage WebSocket and REST connections, making it portable across Linux, macOS, and cloud platforms without language-specific SDKs required.

Streaming STT supports live microphone input, telephony feeds, and continuous audio streams via WebSocket. Each source streams audio frames to AssemblyAI, returning transcript events for meeting notetakers, voice-activated applications, and call-center monitoring.