HydrogenAudio

Hydrogenaudio Forum => Listening Tests => Topic started by: BearcatSandor on 2017-02-16 01:48:39

Title: no audiable differences between audio libraries, right?
Post by: BearcatSandor on 2017-02-16 01:48:39
As a Linux user my box makes use of gstreamer and ffmpeg.  Assuming that they are set up the same way, playing the same file, and using the same soundcard is there any reason to think that these libraries themselves might produce a different [objective] quality of sound?

Almost certainly "no", correct?
Title: Re: no audiable differences between audio libraries, right?
Post by: kode54 on 2017-02-16 01:59:09
Anything which makes use of gstreamer is incapable of gapless decoding of lossy formats, and may very well be incapable of gapless playback of lossless formats. It's like DirectShow, only slightly worse.
Title: Re: no audiable differences between audio libraries, right?
Post by: lazka on 2017-02-16 11:16:42
Anything which makes use of gstreamer is incapable of gapless decoding of lossy formats, and may very well be incapable of gapless playback of lossless formats. It's like DirectShow, only slightly worse.

Depends on the format, Vorbis/Opus should be gapless.
Title: Re: no audiable differences between audio libraries, right?
Post by: [JAZ] on 2017-02-16 20:38:07
I'm not even sure I understand what you are trying to compare.

gstreamer is some sort of audio-video "router". It has plugins which link to other native libraries in order to decode the content, can apply some processing (so, the codec support can be changed without changing gstreamer), and generally it is used by another application to show/play the content to the user.

ffmpeg is an audio/video library, that can encode and decode multiple formats and containers, usually with its own codecs (so, not depending on other libraries, although it is capable of using them) and is generally used alone, although it can be used by other applications as a decoder/encoder too.

So by concept and by implementation, they are quite different, and there is a difference in codec support. As such, the "keeping everything equal" part is quite difficult to get.

Decoding a file should not be a process that generates different output depending on the implementation, but in practice it is for many formats. The most basic ones (like playing a WAV, or even decoding a DPCM signal) should always generate the same bits, but for other formats like MP3, the precision of the internals of the decoder and some of the decisions about the standard can generate differences, which should not be audible, except if they have bugs.
Many years ago, libmad was all on the rise for its "24bit output" as opposed to the 16bit output of many of the decoders of that time. But other than higher precision, the audible difference was questionable.
Title: Re: no audiable differences between audio libraries, right?
Post by: BearcatSandor on 2017-02-16 23:31:07
Thank you [JAZ]. That answers my question succinctly.