AssemblyAI Speech-to-Text API

What it does

AssemblyAI Speech-to-Text API converts recorded audio and live audio streams into transcripts. Developers submit an audio file or connect a streaming session, then receive text with supported timing, confidence and speaker information. The chosen model determines language coverage, recognition behavior and available options such as key-term prompting or speaker labeling.

Teams use the service to build meeting tools, call-review systems, captions and voice agents. Recorded-file workflows process an existing recording, while streaming workflows deliver text during a live interaction. Optional transcript analysis, redaction and specialist features can add useful structure, but their availability and charges differ from basic transcription. Applications must handle storage, permissions and how the transcript is presented to users.

The API suits developers integrating speech recognition into their own products. Create an account, obtain an API key and use an HTTP request, streaming connection or supported SDK. Billing depends on the model, audio duration, channels and add-ons. Evaluate recognition with representative speakers and vocabulary, and check concurrency and usage limits before moving a test into production.

What it can do

• Transcribe recorded audio files
• Receive live transcripts over a streaming connection
• Add supported speaker labels and timestamps
• Select models for the required languages and recognition task
• Use supported key-term and vocabulary controls
• Apply optional transcript analysis or redaction with separate entitlement and charges

Works with

• LiveKit
• Pipecat
• Twilio
• Langflow

Example workflows

Transcribe a recorded call with speaker labels

Illustrative example. Input: a recorded audio file. Steps: send the file with speaker_labels enabled and language detection on. Output: a transcript with speaker-labeled utterances and timestamps.

Add a live transcript to a voice agent

Illustrative example. Input: a live audio stream from a call. Steps: open a Streaming Speech-to-Text WebSocket session and forward audio chunks. Output: partial and final transcript text that the agent can act on.

Before you choose

• Language coverage and available features depend on the selected recognition model.
• Multichannel audio and optional processing can add charges.
• Test recognition, latency and speaker attribution with representative recordings.

Listed under

• Coding
• API
• Audio
• Call analytics
• Developers
• Meeting notes
• Speaker diarization

Get started

Create an account, get the API key from the Dashboard, install an SDK, and call transcribe with audio.

1. Create an account at the AssemblyAI dashboard and copy the API key.
2. Install the Python or TypeScript SDK, or call the HTTP API directly.
3. Initialize the client with the API key and submit an audio file or stream to the transcription endpoint.

Speech recognition is billed by audio usage with model-specific rates. Multichannel processing and add-ons can add charges. Confirm the starting free allowance and whether it recurs; separate text-to-speech prices do not apply to this listing.

Source: assemblyai.com/pricing, checked 29 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Pre-recorded Speech-to-Text API | AssemblyAI ↗www.assemblyai.com
• AssemblyAI | Pricing | Production-grade AI Models ↗www.assemblyai.com
• AssemblyAI Documentation ↗www.assemblyai.com

Company website

assemblyai.com: product, pricing and policy pages

YC directory mirror

Batch and company name

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.