Loss of precision is often a a result of floating point calculations. That is why I mentioned it. The calculations used to compress and decompress files is entirely an integer process. The program may make mistakes, but given you can convert to wav to flac and back to wav and have a bit identical file says that the conversions are being done properly. There is no evidence of any program error or loss of precision. The files before and after conversions are the same based on binary compares, not just checksums.
As to bytes versus integer, in a typical 16/44 file each datapoint is an 16 bit number, which cannot be represented by a byte. Hence, my reference to integers rather than bytes. But the real point is that the calculations do not result in any loss of precision.
The wav to flac to wav tests are just that - tests. They try to see if the conversions influence the sound. People have done this because of the reasons I stated. Some people believe the original and final wav files do sound different, even though the original and final wav file are bit for bit identical. You can say STOP, but people report that they can sound different. I am just summarizing what some people report.
As to bytes versus integer, in a typical 16/44 file each datapoint is an 16 bit number, which cannot be represented by a byte. Hence, my reference to integers rather than bytes. But the real point is that the calculations do not result in any loss of precision.
The wav to flac to wav tests are just that - tests. They try to see if the conversions influence the sound. People have done this because of the reasons I stated. Some people believe the original and final wav files do sound different, even though the original and final wav file are bit for bit identical. You can say STOP, but people report that they can sound different. I am just summarizing what some people report.