How to settle a word you cannot make out in a noisy recording
Tell Vizard Agent which two words it might be. It cleans the track gently, because a heavy denoise removes the consonants that distinguish them. Then it re-transcribes a longer stretch so the surrounding sentence constrains the answer, and compares the two candidates directly rather than transcribing the phrase again from scratch.
What do you type into Vizard Agent?
Give Vizard Agent both words. The instinct is to ask what the word is, but that is the question the first transcript already got wrong, and asking it again in the same open form tends to produce the same wrong answer with the same confidence. Two candidates turn an open question into a comparison, and a comparison can be settled.
Prompt
What does Vizard Agent actually do?
Noise reduction removes what does not look like speech, and the quiet consonants at the ends of words are exactly what it mistakes for noise. So the clean is set for how well the audio transcribes, not how clean it sounds. A word heard alone is a coin toss, while a word inside a sentence is constrained by grammar and sense. In the session behind this article it:
- Transcribed the speech with timings, re-listened to the short phrase and confirmed the language.
- Cleaned the background noise off the track, re-transcribed it and listened to narrow the disputed word. That first clean was too aggressive.
- Redid the clean gently, boosting the voice instead, and re-transcribed after the gentler processing.
- Took a longer stretch of the original, cleaned it gently and transcribed it again, so the surrounding sentence could constrain the word.
- Compared the two candidate words directly on the cleaned audio.
When does this not work well?
Some words are simply not recoverable. If the noise sits at the same frequencies as the speech, the audio is clipped, or the speaker mumbled straight through the word, no processing will put back information the microphone never captured. A heavily compressed upload may have lost the detail too. Vizard Agent tells you so rather than picking one.
Common questions
Do I have to give two options? No, but it helps a great deal. Vizard Agent can compare candidates directly.
Will it guess if it cannot tell? No. Vizard Agent says when the audio supports neither candidate, which is more useful than a confident wrong word.
What if it is a name? Tell Vizard Agent it is a name. That changes which spellings are plausible.