Free credits
$50one-time credit for new accountsCovers Pre-recorded STT, Real-time STT, Speech Understanding and Guardrails; credits do not expire. LLM Gateway is excluded. Official source 1 ↗
Website-checked tool profile
AssemblyAI
Voice AI APIs for speech-to-text, dictation, text-to-speech and speech understanding, billed per hour of audio.
Sources checked
- Pricing
- Pay-as-you-go with no contracts, minimums or subscriptions, priced per hour of audio by model; new accounts get $50 in free credits. Pre-recorded transcription is $0.21/hr with Universal-3.5 Pro or $0.15/hr with Universal-2, realtime from $0.15/hr, Sync $0.30/hr and Dictation $0.62/hr; add-ons such as diarization are billed separately. Custom rates are available from sales. Official source 1 ↗ · Official source 2 ↗
- Available as
- Cloud API (HTTP and WebSocket) with Python and TypeScript SDKs, an MCP server for coding agents such as Claude Code and Cursor, integrations with LiveKit, Pipecat and Twilio, a self-hosted option, and purchase through AWS Marketplace. Official source 1 ↗ · Official source 2 ↗
- Primary group
- Voice & Speech
Also consider
See all 6 similar tools ↓- Zen WhisperLocal-first Mac dictation and transcription that types into your existing apps.Free plan
- Muse Voice TranscribeMeta's speech-to-text API model for live and recorded audio, with speaker labels, turn detection and 25 languages at $0.18 per hour.Paid
- TellPipA Mac call assistant that suggests what to say from your documents and client memory, with local transcription.Free trial
What it does
AssemblyAI is a developer platform of voice AI APIs: pre-recorded, streaming and synchronous speech-to-text, a dictation API, text-to-speech for voice agents, speech understanding, guardrails and an LLM gateway. Developers call it through the HTTP and WebSocket API or the Python and TypeScript SDKs, on AssemblyAI's cloud or self-hosted.
Official source 1 ↗ · Official source 2 ↗Features
- Pre-recorded speech-to-text. Asynchronous transcription of audio and video files across 99 languages, with speaker diarization, PII redaction, word-level timestamps, keyterms prompting and custom spelling. Official source 1 ↗ · Official source 2 ↗
- Realtime streaming transcription. Live audio over WebSocket at about 150 ms p50 latency for voice agents, live captions and agent assist, with end-of-turn detection. Official source 1 ↗ · Official source 2 ↗
- Sync and Dictation APIs. The Sync API returns a finished transcript in one HTTP call without polling; the Dictation API also cleans up and formats the text with an LLM per your instructions. Official source 1 ↗ · Official source 2 ↗
- Text-to-speech for voice agents. AssemblyAI-TTS-1 streams synthesized audio back, with thousands of voices and voice cloning in six languages. Official source 1 ↗
- Speech understanding and guardrails. Summarization, sentiment, entity and topic detection on transcripts, plus PII redaction and content moderation, enabled on the same transcription request. Official source 1 ↗
- Medical mode and LLM Gateway. An add-on Medical Mode improves accuracy on clinical terminology, and the LLM Gateway gives access to models such as Claude, GPT and Gemini through the same API, billed per token. Official source 1 ↗ · Official source 2 ↗
Pricing & free limits
Pre-recorded: Universal-3.5 Pro
$0.21per audio hour, pay as you go18 languages with code switching; add-ons from $0.02/hr (diarization) to $0.15/hr (Medical Mode). Official source 1 ↗
Pre-recorded: Universal-2
$0.15per audio hour, pay as you go99 languages; keyterms prompting included. Official source 1 ↗
Realtime: Universal-3.6 Pro Realtime
$0.45per streaming hour, pay as you go32 languages, context carryover, self-correcting speaker labels and keyterm prompting. Official source 1 ↗
Realtime: Universal-Streaming
$0.15per streaming hour, pay as you goEnglish-only, or the Multilingual variant for English, Spanish, German, French, Portuguese and Italian at the same rate. Official source 1 ↗
Sync API
$0.30per audio hour, pay as you goUp to 2 minutes per request with results in about 134 ms (p50). Official source 1 ↗
Dictation API
$0.62per audio hour, pay as you goFormatted text from one utterance, 32 languages, clips up to 120 seconds. Official source 1 ↗
Custom
Customcontact salesCustom rate limits, higher concurrency and volume discounts. Official source 1 ↗ · Official source 2 ↗
Billing details
You prepay funds by card and usage is drawn down; pre-recorded audio is pro-rated to the second and failed requests are not charged. Streaming is billed on how long the WebSocket stays open, including idle time, and unclosed sessions bill for up to 3 hours. Auto-pay can top up the balance; SSO is a $199 monthly add-on per connection. Official source 1 ↗
Free access & limits
$50 of non-expiring free credits for new accounts, excluding LLM Gateway. After that a card is required, and API access pauses if the balance reaches $0 without auto-pay. Official source 1 ↗
Best for & limitations
Build voice agents that transcribe callers in real time and reply with low-latency speech. Official source 1 ↗
Power meeting notetakers and call analytics that transcribe, summarize and analyse recorded calls, podcasts or meetings at scale. Official source 1 ↗ · Official source 2 ↗
Add medical transcription or ambient clinical-note features to healthcare products. Official source 1 ↗
- Streaming bills idle time. Streaming is charged for the whole open session, including idle time; sessions left open auto-close after 3 hours and bill the full 3 hours. Official source 1 ↗
- Sync and Dictation clip length. Sync and Dictation requests handle up to 20 minutes over HTTP, but only up to 120 seconds through the SDKs or when uploading while recording. Official source 1 ↗
- Model language coverage varies. Universal-3.5 Pro covers 18 languages and Universal-Streaming is English-only; 99-language coverage applies to Universal-2. Official source 1 ↗
Similar tools
Tools like AssemblyAI
Published tools that share its categories.
Zen WhisperLocal-first Mac dictation and transcription that types into your existing apps.
Visit Muse Voice TranscribeMeta's speech-to-text API model for live and recorded audio, with speaker labels, turn detection and 25 languages at $0.18 per hour.
Visit TellPipA Mac call assistant that suggests what to say from your documents and client memory, with local transcription.
Visit 
VowiseVoice dictation and transcription app that turns spoken notes into a journal, reviews and AI memory.
Visit 
Audio8 ASREdge0's open speech-to-text models: live 24/7 Chinese/English streaming, plus a 0.1B model for CPUs and iPhones.
Visit 
CluinnAI dictation for any Mac app, plus a call assistant that answers questions mid-call and lists action items after.
Visit