Asset uPnP Audio quality settings

[mrbadmf]

Using Asset on Qnap NAS.
Marvelously working server for network streaming. And GAPLESS!!! The only ones which serving it right through my AV receiver Pioneer SC-LX801 are Asset uPnP and Plex. Asset is winner by a mile because of its user friendliness and much better sorting options.
Just wondering if there will be anything done for audio quality settings/upscaling.
Or are audio files always played through in original specs?
Can't find anything on this, so yeah it's a mystery to me...

Thanx again for this wonderful package!

Kind regards, Luc

[Spoon]

Re: Asset uPnP Audio quality settings

You can set to transcode the audio to wave, and at the same time upscale to 24 bit if so wished.

[timster67]

Re: Asset uPnP Audio quality settings

You can set to transcode the audio to wave, and at the same time upscale to 24 bit if so wished.

Really? According to the manual, the upscaling is only used as part of applying ReplayGain tags:

Apply ReplayGain ReplayGain is a method of measuring loudness of audio tracks and altering the overall loudness, enabling tracks from different albums to have the same volume levels (modern popular albums are much louder than an album from the 1980's). ReplayGain requires that the audio tracks have an embedded ReplayGain ID Tag (dBpoweramp is able to add this tag to files, install dBpoweramp and the ReplayGain utility codec from Codec Central on dbpoweramp.com, then run the program 'Batch Converter', select the main audio folder (normally Music) and select Convert >> choose the encoder the [ReplayGain]. This method only scans the files and adds ReplayGain values, the audio is left untouched).
Only tracks which are streamed (in a format other than 'as is') would have replaygain applied. If the player supports 24 bit audio then enable this option as it allows ReplayGain adjustments without any loss of bit depth when playing 16 bit tracks, also set the Active and Inactive pre-amp to -3 dB.

[Spoon]

Re: Asset uPnP Audio quality settings

If you do not have replaygain tags then no volume changes are made.

[timster67]

Re: Asset uPnP Audio quality settings

Ahh, so you can tick the boxes and if no tags are present it will still upsample? Good to know, but perhaps the manual could say that? It seems a bit of a fudge - would it not be possible to be able to tick the 24-bit box and not the ReplayGain one?

[misha0209]

reviving this old thread as i have been searching for this exact same thing the other day.

so i agree with what is posted here, i do not think it is a good idea having the bit depth rescaling linked to replay gain. it would be much better to have something like WAV WAV16 WAV24 and WAV32 (why not) when you choose how to transcode each format. making the user enable replay gain to get 24 bit depth in all cases, means i need to remove any replay gain tags i might have on my songs (maybe i downloaded an album, and do not know they are there) to be sure there is no applied volume.
also if users follow the manual and "also set the Active and Inactive pre-amp to -3 dB." they are implicitly applying volume change to all tracks... not nice...

also i would like to ask a question. how does asset play? does it play from a RAM buffer, or serve from disk?

why i ask, i use asset on a respberry pi, with usb storage, that can have latency.
if i transcode to WAV, there is a clear improvement, i assume this is attributed to 2 factors:
1. streamer does not decode flac/mp3/whatever, it is already decoded by asset, and presented as WAV (here i would like to experiment with WAV24 as well).
2. if transcoded to WAV, asset for sure serves from a buffer in RAM, because it needs to decode, and it would make no sense to write the decoded file.
as such, asset is getting rid of any delay the USB drive of the PI might have during playback. which is GREAT !!!

now, if i play DSD files, i do not want them transcoded to PCM, as my streamer has DSD support, so i present them as is.
but in this use case, how does asset serve the file? read and serve from disk at playback? or does it load a part of the file to RAM, and it serves from there? (just like when transcoding but without transcoding)
it would be nice if it would serve from RAM buffer. if not, can this be done?
it would also be nice if user could adjust the buffer (i am aware that setting too high a buffer will result in a delay when playback starts, as asset would be limited by the USB drive speed).

[Spoon]

The player does the buffering, there is effectively no latency otherwise you would hear audio breaking up.

[misha0209]

thanks for the reply.
i realized after writing that i used the incorrect term i did not mean delay, as in delay in serving data.
"asset is getting rid of any delay the USB drive of the PI might" i mean to say that any latency or jitter introduced by serving files directly from disk is nullified by buffering to RAM and serving from there. provided buffer is big enough, samples are sent EXACTLY when intended.
good to hear that dsd files are also buffered to RAM and pushed to streamer from RAM !

regarding the other points, 24 bit being done not via replay gain, but maybe added as options in the presentation fields of each file type? is that something that can be done?
also, is it possible to configure the size of the buffer to be loaded to RAM? both with and without transcoding? i use an 8gb rpi4b, so i have plenty of ram commit as buffer.

[Eric Seaberg]

Let me mention that if you're up-scaling from 16-bit to 24-bit, you're NOT gaining anything other than 8-zeros in the bit stream... and a larger file. The quality isn't improving at all because you're starting with 16-bit from the get go, and it can NEVER get better than that.

Same thing with up-scaling from a sampling rate of 48KHz to 96KHz or higher... there's nothing in the original data stream to make it better. You're just making a larger file of the same thing.

[misha0209]

yeah, my logic dictates the same thing... there's no more in formation in the original lossless file (or a decoded lossy file), and just padding with 0 will not bring anything new. unless you do something like DSP.. or volume adjustment, or resampling, so change the source in some way, then the extra headroom can prevent clipping for example.

it is just that i read on multiple forums that some streamers might handle incoming data differently if presented with a 16 or 24 bit depth, or might prefer a specific incoming sample rate.
i have experimented with the replay gain to get 24bit and yes, i perceive no difference, as i expected.
so maybe it is useful for a dodgy streamer/dac implementation? i don't know. hence it still might be a useful feature to have.
in my case though, it changes nothing.
i wanted to try the 24bit thing, because asset brought a huge improvement in my setup, and it got me wondering if rumors regarding bit depth might also be true, even if logically i knew it made no sense ))

i believe the right way to go is to use server side to do decoding and present uncompressed file of the same bit depth and sample rate as original to the streamer.

however i still say that if some users wish to upscale to 24 bit (just add 0), this should be intuitive, straight forward and not linked to the replay gain (even if this is the only use case where it makes any sense).