Unsure which audio format for voice biometrics?
- 8-bit A-Law
- 8-bit Mu-Law
- 16-bit linear PCM (little-endian or big-endian, depending on your architecture)
You can also submit audio data as “raw” binary, i.e. without a header/container, in which case you need to specify the actual “SampleFormat” of the binary data when you call SubmitPhrase() on our APIs. Note that other formats, such as MP3 and MP4, or different frequency audio samples will be rejected by the voice biometric engine.
If there is a mismatch between the audio format that was specified to that which was submitted you will receive a
SampleFormatMismatch error from
If the audio file was specified but does not contain enough information to help determine what the format actually is, you will receive a
SampleFormatUnsupported error message from
SubmitPhrase. For example, “Wave” could be specified and valid PCM audio files are submitted but without RIFF headers, so that we are unable to determine whether the file is big-endian or little-endian.
Finally, if you are using the REST API then you may receive an ‘Error delegating to SOAP service’ response when the audio sample does not match the required format.