AssemblyAI API

AssemblyAI API provides enterprise-grade speech-to-text + LLM enhancements — real-time and batch transcription, speaker diarization, sentiment analysis, auto LLM summary.

Visit site ↗Documentation ↗Health checked 9h ago

Use it when

Industry-leading transcription accuracy (best WER 8% English)

Watch for

Paid product (from $0.37/hour)

First check

Sign up at assemblyai.com for API key. POST /v2/transcript with audio URL → get transcript ID → GET /v2/transcript/{id} for results.

Auth

api_key

CORS

HTTPS

Yes

Signup

Latency

492 ms

Protocol

REST, WebSocket

Pricing

paid

Uptime · 30-day window

Probes: 1Uptime: 100%Avg latency: 492ms

About this API

AssemblyAI is a speech AI company founded 2017, focused on speech-to-text + downstream NLP. Unlike general LLM companies (OpenAI Whisper), AssemblyAI is vertically deep: in-house Universal-2 model leads English WER (Word Error Rate); provides LLM enhancements (auto summary, entity extraction, custom topic detection); accurate speaker diarization (who said what in a meeting auto-tagged). vs. OpenAI Whisper: AssemblyAI is paid but quality more stable, enterprise SLA better; Whisper is open-source free but requires self-run GPU and no diarization. Customers mainly in meeting tools (Otter, Zoom), call center quality platforms, media companies (subtitle generation). Weak Chinese support is a known limitation; for Chinese transcription, recommend self-run Whisper or iFlytek / Alibaba Cloud Speech.

What you can build

1Auto-transcription of meetings (Zoom, Google Meet integration)
2Call center quality analysis
3Podcast / YouTube subtitle generation
4Medical / legal / financial transcription

Strengths & limitations

Strengths

Industry-leading transcription accuracy (best WER 8% English)
Low-latency real-time streaming
LLM enhancements (summary, entity extraction)
Accurate speaker diarization

Limitations

Paid product (from $0.37/hour)
Weaker Chinese support (focused on English + major European languages)
No free tier (short trial credits)

Example request

Generic template — replace <endpoint> with the real path from the docs.

curl https://www.assemblyai.com/<endpoint> \
  -H "Authorization: Bearer $API_KEY"
# Some providers use X-Api-Key instead — verify in the docs.

Getting started

Sign up at assemblyai.com for API key. POST /v2/transcript with audio URL → get transcript ID → GET /v2/transcript/{id} for results.

FAQ

AssemblyAI vs. Whisper?+

Need enterprise SLA + diarization: AssemblyAI (paid). Open-source + self-run GPU: Whisper (cost-effective).

What about Chinese?+

Not recommended for Chinese with AssemblyAI. Whisper handles Chinese reasonably; domestic iFlytek and Alibaba Cloud Speech are best-optimized for Chinese.

Technical details

CORS: ?HTTPS: YesSignup: ?Open source: No

Auth type: api_key
Pricing: paid
Rate limit: pay-as-you-go 无明确 RPM 限制
Protocols: REST, WebSocket
SDKs: python, javascript, typescript, go, java
Response time: 492 ms
Last health check: 5/12/2026, 7:36:55 AM