My audio codecs benchmark

2020-12-26 22:09:47

Hi Guys,

I'm new to the forum, but I have been using fb2k and collecting albums for years. Recently I just want to confirm that I chose the right codec for my collection (I use wavpack), so I created my benchmark. The link is https://github.com/cmpute/audio-codec-benchmark, it's open and feel free to run your own benchmark with different codecs and audio sources!

I compared the

• compression ratio, coding time and some criteria for audio quality of the lossy codecs.
• compression ratio and encoding/decoding time of the lossless codecs.

Suggestions for better and fairer comparisons are welcome (comments here or issue ticket on Github)



Mod edit: Merged topics discussing the same benchmark script.

Re: My audio codecs benchmark

Reply #1 – 2020-12-26 22:44:49

Nice, thanks. Although spectrum diff and SNR for comparing lossy codecs might get some beating here

Re: My audio codecs benchmark

Reply #2 – 2020-12-27 04:58:49

Yeah, the SNR and spectrogram measurements are nonsense.

Re: My audio codecs benchmark

Reply #3 – 2020-12-27 06:43:23

Same. I was triggered
Maybe you should rather gather data of all the known listening tests and base the audio quality graphs off of those. Those SNR and spectrogram measurements will give unexperienced newcomers the wrong idea of which lossy encoders perform better.
You can also include exhale in your encoder list for xHE-AAC.