Speaking Lab

Listen, record, listen back — then try one useful change.

Enter a reference sentence to practise

Listen for meaning and pauses, read it aloud, then choose one thing to improve after listening back.

Voice settings & listening practice

Model accent Speed

These settings apply to lesson playback in this browser. If the accent is unavailable, another English voice is used.

Record and listen back

Record up to 60 seconds and retry as often as you like. A new take replaces the current recording; your chosen best take is stored separately.

0s / 60s

You choose your best take after listening; this is not an AI score. Audio is stored in this browser and is deleted if you clear its data.

Optional: transcription and feedback

Browser recognition may send live speech to your browser provider. Whisper / Azure send audio to their respective services. AI feedback sends text only. Unconfigured services do not generate results.

I consent to using the selected speech or AI service.Optional OpenAI API key (page memory only)

Requests using your key are billed to your OpenAI account. Reloading clears the key.

Transcript or typed response (check recognition errors)

Reading pronunciation · up to 30 seconds

Could you send me the updated quote by three?

Record the reference above before requesting a score. Azure evaluates this reading, not business judgement or overall English proficiency.