Speech, word by word
Record your voice, or upload a WAV, and read the transcript while you talk; every recognized word keeps its own start and end time.
Your microphone stays closed until you press Record, and loading this page makes no call to the speech provider.
Three sailors watched the ship carry rice past the pier.
US English
PCM16 WAV, 1–30 s, mono/stereo, at 16, 44.1 or 48 kHz. Converted automatically.
Recognition is unavailable on this deployment. You can retain and replay audio; full assessment requires recognition.
Status: Ready
What we heard
- Nothing recognized yet.
Speech assessment
Results compare the recording with the expected text and pronunciation. Word playback times are approximate.
Pronunciation assessment is unavailable on this deployment.
Synthetic Speechify example · acoustically unverified
Pause the example before recording. This is not verified perfect pronunciation or an assessment label.
sailors-geffen_32-1 · speechify-example-v1 · SHA-256 5fe34c8936c46de41e207a75d9475a037574681e0e4a39f81efda458c5d3c8d2
How to record
- 1
Allow the microphone once
The browser asks for permission the first time you press Record, and nothing is captured before that. Loading this page makes no call to the speech provider.
- 2
Speak, then stop
Partial text shows up while you talk. The attempt stops on its own at 30 seconds; Cancel releases the microphone right away and skips the final transcript.
- 3
Check the transcript and the timings
Final sentences and word timings land in the list above. Every row replays the recording from that word, so you can hear what the model heard.
Provider diagnostics
- Profile
- qwencloud
- Model
- qwen-audio-3.1-asr-flash-streaming
- Region
- shared
- Deployment scope
- shared
- Endpoint
- wss://maas.qwencloudapi.com/api-ws/v1/inference
- API key
- configured
- Receipt signing
- configured
- The speech relay is limited to administrator accounts on this deployment.