Has anyone else noticed how the buffer setting under playback option in Volumio affect sonic character?
I run the latest compatible free version of Volumio with a Raspberry 5 8GB and use the USB output to feed a Topping D90III Discrete DAC. High Resolution music files are fed from an attached HDD. My observation is as follows:-
1. With the buffer setting at 12MB the sound is perceivably more impactful. But this impact is because of everything in the music becoming more apparently defined but actually a bit strident. Sound stage increases in width and height. Good recording sound good whereas bad recordings are unlistenable.
2. With the buffer set to 1MB the sound stage decreases in width and height but increases in depth. The music is more soulful, analog, better timed and very relaxed. The stridency simply disappears and even bad recordings are just about enjoyable. Good recordings are phenomenal, reminding me of the good old analog days. Low level resolution is excellent.
Any thoughts on this? Thanks in advance.
I run the latest compatible free version of Volumio with a Raspberry 5 8GB and use the USB output to feed a Topping D90III Discrete DAC. High Resolution music files are fed from an attached HDD. My observation is as follows:-
1. With the buffer setting at 12MB the sound is perceivably more impactful. But this impact is because of everything in the music becoming more apparently defined but actually a bit strident. Sound stage increases in width and height. Good recording sound good whereas bad recordings are unlistenable.
2. With the buffer set to 1MB the sound stage decreases in width and height but increases in depth. The music is more soulful, analog, better timed and very relaxed. The stridency simply disappears and even bad recordings are just about enjoyable. Good recordings are phenomenal, reminding me of the good old analog days. Low level resolution is excellent.
Any thoughts on this? Thanks in advance.
If buffer size has such an audible impact then something is obviously wrong with Volumio. Larger buffer size may typically help with dropouts if the HW is not up to the task but the audible impact you describe is not related to dropouts.
IIUC that "Audio Buffer Size" config in volumio configures audio_buffer_size parameter of mpd https://linux.die.net/man/5/mpd.conf. Looking at mpd source code - that parameter sets size of internal buffer for decoded samples. This buffer contains chunks (approx 4kB) of audio data plus chunk metadata (used for mixing the chunks, gain setting etc.). Data from this buffer are passed to the output device in a separate thread which breaks these chunks into device period size (Alsa Output Parameters - buffer_time and period_time). No data are actually passed inside MPD buffer as the chunks are just pointer structures.
The buffer size parameter works in combination with the buffer_before_play parameter to allow continous playback on sources which provided data in large chunks.
A double blind listening test would be required to confirm any audible effect of length of this large internal buffer first as that is extremely unlikely.
The buffer size parameter works in combination with the buffer_before_play parameter to allow continous playback on sources which provided data in large chunks.
A double blind listening test would be required to confirm any audible effect of length of this large internal buffer first as that is extremely unlikely.
