Technique for Academic Speaking • short article presentations • strength–limitation analysis
Tasks 1–4 use short model scripts and allow up to 45 seconds for each speaking attempt. Task 5 lets you paste your own English presentation script (maximum 135 words) and speak for up to 70 seconds.
The automatic score uses speech recognition as an intelligibility proxy, comparing the recognised speech with the script. TTS provides the model voice; TTS itself does not score pronunciation.
Model listening and speaking attempts are unlimited. Practise as many times as needed; only the highest score for each task is kept for the final result.
Current Chrome or Edge is recommended. Current Safari/iPadOS is supported when browser speech recognition is available and microphone permission is granted. Leaving the window or using split view does not lock the lab; the event is warned and recorded.