Google's latest generative ASR foundation model — 85+ languages
68 languages · benchmarked
Chirp 3 is a speech-to-text model from Google Cloud. In our standardized benchmarks it reaches 9.95% word error rate, ranking #4 of 35 models tested. It supports 68 languages and runs at $0.011/min — a premium option.
Its nearest benchmarked alternative is Soniox STT (Async). Best suited for multi-speaker conversations, multilingual workloads, and accuracy-critical work.
Overall Score
45.0
#4 of 35
Word Error Rate
9.95%
Character Error Rate
5.81%
Match Error Rate
9.65%
Word Info Lost
14.11%
Avg Latency
11.0s
Benchmarks Run
34
last 30 runs
Word error rate
9.95%
↓ 29.2% · 30d
3 runs ago
latest
Avg latency
11.0s
↑ 7.1% · 30d
3 runs ago
latest
overall WER · 35 models
Chirp 3
field · 34 models
lower-left is better
8 categories
WER
lower is better
Medical
1.9%
Legal
2.9%
Technical
3.7%
Conversational
4.7%
Finance
7.1%
General
8.2%
Code-Switching
17.1%
Noisy Environment
29.2%
5 accents
Rate
$0.0107
/min
Per-second billing. Bring your own provider key and pay your provider directly — a 5% routing fee applies (first 100 min/mo free).
Set up BYOKCost estimator
1 hour
$0.64
10 hours
$6.41
100 hours
$64.08
1,000 hours
$640.80
Billed per second of audio processed.
Model ID
google/chirp-3
Authenticate every request with your secret API key as a Bearer token. Issue a key from your dashboard.
Use it with your agent
Paste this into Claude Code, Cursor, Codex or any agent with a shell. It installs the CLI and the OpenTranscription skill, then transcribes with this model.
Quickstart
Upload your audio, create a job with this model, then poll the job or set a webhook_url to be notified when it completes.
Key parameters
file_path
string
required
Storage path returned by the upload step.
model
string
required
The model to run this job on.
models
string[]
Ordered fallback chain: primary first, then backups tried in order if a provider fails.
language
string
ISO 639-1 language code. Omit to auto-detect.
diarization
boolean
Label speakers (A, B, …) in the output.
custom_words
string[]
Names, jargon and product terms to bias the model toward. Up to 1000 entries.
vocabulary_list_id
uuid
A vocabulary list saved in your workspace. Merged with custom_words when both are sent.
webhook_url
string
HTTPS URL notified with the result when the job completes. Delivered events are signed (X-OT-Signature).
title
string
Display name for this job in the dashboard. Falls back to the file name.
Response
A completed job returns the transcript in the OpenTranscription Unified Schema (OTUS).
Webhooks — skip polling
We POST a signed event to your webhook_url when a job completes or fails; verify the X-OT-Signature (HMAC-SHA256, reject if older than 300 s) and dedupe on the event id, then fetch the full transcript via GET /api/v1/transcriptions/{id}.
Rate-limited per tier — see the X-RateLimit-* response headers.
Full API referenceUptime · 30d
66.67%
Error rate
33.33%
Avg latency
108.6s
Last incident
None in 30d
from the same field
same benchmark, every match
Supported Languages
Supported Formats
Features
Limits
Max file size
No documented limit
Max duration
8 h