result and segments are empty

[illusion77]

For this code

const filePath = `${RNFS.DocumentDirectoryPath}/whisper/ggml-tiny.en.bin`
const sampleFilePath = `${RNFS.DocumentDirectoryPath}/youtube_audio_3LopI4YeC4I.wav`

const whisperContext = await initWhisper({ filePath })

const options = { language: 'en' }
const { stop, promise } = whisperContext.transcribe(sampleFilePath, options)

const { result, segments } = await promise
// result: (The inference text result from audio file)

console.log('---result---', result);
console.log('---segments---',segments);

I got this print

---result---
---segments--- Array []

This audio file is a English content, and the pronunciation is pretty clear
I do not know why got this.

Thanks!
@jhen0409

[illusion77]

I found that the count might be normal

But the n_segments are 0, this is wrong

Actually, I met the same situation (n_segments = 0) when I use whisper.cpp.swift, it is fixed after I configured 'Copy Bundle Resources', but I do not why for this

[jhen0409]

You can check the WAV file is 16khz mono first, only this format is currently supported in whisper.cpp.

Otherwise, you can also try to find related issues in whisper.cpp repo.

[illusion77]

Thanks, my process is to download the audio file in webm format, then use ffmpeg-kit-react-native to convert it into the same metadata as jfk.wav, it still doesn't work...