JackMostow
commented
2 years ago [ ] extract Chirag’s code in narrate_mode branch for writing input audio utterances to files [ ] generalize to work for any activity that listens, not just for narrate_mode [ ] add config.json switch to record utterances (just for activities that listen) [ ] extend to record sentences instead, if feasible [ ] store not just the audio, but also:
- Expected word sequence (e.g. displayed sentence, minus punctuation)
- Expected starting position in text (location of first word to expect)
- Final ASR output hypothesis
- Sequence of partial ASR hypotheses (if feasible)
Is your feature request related to a problem? Please describe. A clear and concise description of what the problem is. Audio recording disables ASR, so we have to turn it off during ASR.
Describe the solution you'd like A clear and concise description of what you want to happen. Record audio input.
Describe alternatives you've considered A clear and concise description of any alternative solutions or features you've considered. We record audio except during ASR.
Approach to be followed (optional) A clear and concise description of approach to be followed. Copy audio input from mic input used in ASR. Octav implemented it for OS 8 but not earlier OS versions. Chirag must have implemented it for NARRATE mode, but I don't know how. See how he did it.
Additional context Add any other context or screenshots about the feature request here.
Record audio input Session / activity to support debugging and usability studies [mostly done] Sentence / utterance to support narration and off-line ASR experiments