Skip to main content

Troubleshooting

Common issues​

  • Problem:
    The result contains a lot of silence predictions even though there is audio
    Resolution:
    Decrease the volume_threshold of your processing job or if you don't care about silence detection at all - set it to volume_threshold=0.0
  • Problem:
    The API returns unexpected results
    Potential resolutions:
    • Ensure that the data you are sending is in a format that is supported by the API. The supported formats are defined here:
      File processing
      Stream processing
    • Try normalizing your data before sending it to the API.
  • Problem: A file larger than the file-processing size limit needs to be processed
    Potential resolutions:
    • If the file exceeds the 15MB limit for direct upload, try the same method using the URL option.
    • If the file exceeds the 1h limit for the URL method, try to use the streaming API to handle arbitrarily long audio streams - see Example usage in the Real-time processing section.
    • Another option would be to split the file into smaller chunks, send multiple requests and combine the results. To split the file, you can use the following sox command:
      sox <INPUT_FILE> -b 16 file_out.wav rate 16k remix 1 trim 0 420 : newfile : restart

If your issue cannot be resolved by the proposed measures above, please reach out to our support at support@oto.ai.