0

My question is rather a matter of interest than need.

I did a hearing test online to find out the maximum frequency, that I can still hear. The result was ~18500Hz. So I concluded the most efficient way to encode my audio files would be with a 37kHz sampling rate. The thinking was, that would allow each frame to be larger in size with the same overall bitrate.

So: opusenc --bitrate 110 --raw-rate 37000 Sample.wav Sample.opus

The conversion went normal, but the resulting file sounded like random white noise and like the high frequencies being cut off.

So I have several questions:

  • Is the thought process in the first abstract correct? Does a 48000 Hz sampling rate on a file with a maximum frequency signal of 21kHz waste frame quality or does opusenc (or any other mean of conversion) recognize that an sample at 42kHz (or 44,1kHz)?

  • I have heard that even 96kHz sampling could me useful to encode e.g. two signals way above our spectrum, that together produce a hearable sound. Is that true and is that the explanation for the awful resulting file?

  • Does opusenc simply not know what to do and puts out that kind of a file instead of error?

Kind regards

Markus

LiveWireBT
  • 28,763
Markus Gratis
  • 697

1 Answers1

2

So I concluded the most efficient way to encode my audio files would be with a 37kHz sampling rate. The thinking was, that would allow each frame to be larger in size with the same overall bitrate.

I'm not quite sure what you mean, codecs use a low-pass filter to achieve what you want.

opusenc --bitrate 110 --raw-rate 37000 Sample.wav Sample.opus

The conversion went normal, but the resulting file sounded like random white noise and like the high frequencies being cut off.

If your sample is not recorded at 37kHz then you are not using the tool correctly, from the manpage:

--raw-rate N – Set sampling rate for raw input (default: 48000)

Opus only supports a 48kHz sampling rate according to the FAQ:

LiveWireBT
  • 28,763