Managing and monitor the several MS Stream Options

[z624391]

[font=Verdana]

I got some questions:

In which log files and log level can we see how and which streamer/resampler and maybe other options, is MinimServer actually using?

How are the options within the stream.transcode and stream.option influencing i.e. combined together? For instance: "convOut=-af aresample=resampler=soxr -ar 192" and "flac:wav24;192".

In streaming land I only see thousands stories telling not more than half the story.
Which is quite typical for "Linux" environments.

This site is a good example: https://ffmpeg.org/ffmpeg-resampler.html...er-Options

[z624391]

Primarily I like to check which and by what the output signal is generated. In that way I know what I am listening to, this will make changes i.e. optimizations more easy to follow and understand.

Otherwise I imagine listening to things that were not happening at all.

[simoncn]

The settings in the stream.transcode property and the convOut option of the stream.options property are combined to pass a command string to ffmpeg. You can see this command string by enabling Debug logging and looking in the minimserver.log file.

Here is an example of this log entry using your example settings:

Code:

ChainImpl: property filetrans.command has final value '"/mnt/app/opt/MinimServer/minimserver/libext/ffmpeg-melco-4.1.dir/ffmpeg" -v 0 -analyzeduration 0 -f flac  -request_sample_fmt flt  -i "/mnt/disk1/share/music/musicfiles/SACDA67524/!Albricias, mortales!.flac" -af aresample=resampler=soxr -ar 192 -vn  -ar 192000 -f s24le -acodec pcm_s24le -sample_fmt s32 pipe:'

This loooks a bit strange because the -ar option appears twice. In this case I think ffmpeg uses the second value 192000.

[z624391]

Thx. I am little bit playing and testing with the 3 stream.xxx fields to find out their functional behaviour:
- using stream.transcode is overruling the bit and sample setting within stream.options.
- as long as the parameters are syntactically valid, no message appear if their interaction is correct. Therefor I put in double items to see what happens (nothing is reported).
- I also installed the last version of ffmpeg QNAP App incl. the libsoxr. Is it now really using the "alternate" sample library? I can't tell.

It's common accepted that 24bit really improve the sound quality. But for upsampling (except professional recordings) many articles on internet have other ideas about that. At the end you just have to listen to well know music to find out.

[simoncn]

The purpose of convOut in stream.options is to specify additional ffmpeg options that cannot be specified in stream.transcode. The convOut setting is checked to make sure it is valid for the ffmpeg command but MinimStreamer does not attempt to check for duplication or conflict between convOut settings and stream.transcode settings.

If you have specified -af aresample=resampler=soxr in convOut and libsoxr is available in the ffmpeg executable that you are using, ffmpeg will use libsoxr for transcoding. I don't know whether ffmpeg reports an error if you have specified this and libsoxr isn't available.

I agree with your comments about bit depth and upsampling. Many DACs do upsampling internally and this can produce better results than having ffmpeg do it. For bit depth, it is often beneficial to use transcoding to expand bit depth to the maximum that the DAC can handle. The only way to be sure is to listen.

[z624391]

To be sure the QNAP QPKG FFMPEG package is the right one and point to the right location ("/share/CACHEDEV1_DATA/.qpkg/ffmpeg"), I checked "/opt/ffmpeg" and also added "/share/CACHEDEV1_DATA/.qpkg/ffmpeg" to the PATH resulting in:

[/share/CACHEDEV1_DATA/.qpkg/ffmpeg] # ffmpeg -version
ffmpeg version 7.0.2-static https://johnvansickle.com/ffmpeg/ Copyright © 2000-2024 the FFmpeg developers
built with gcc 8 (Debian 8.3.0-6)
configuration: --enable-gpl --enable-version3 --enable-static --disable-debug --disable-ffplay --disable-indev=sndio --disable-outdev=sndio --cc=gcc --enable-fontconfig --enable-frei0r --enable-gnutls --enable-gmp --enable-libgme --enable-gray --enable-libaom --enable-libfribidi --enable-libass --enable-libvmaf --enable-libfreetype --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-librubberband --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libvorbis --enable-libopus --enable-libtheora --enable-libvidstab --enable-libvo-amrwbenc --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libdav1d --enable-libxvid --enable-libzvbi --enable-libzimg

Now I know how it is functionally installed and that it is really using libsoxr.

The original QNAP imbedded ffmpeg is located here: "/usr/bin/ffmpeg" and very ancient:

[/usr/bin] # ./ffmpeg -version
ffmpeg version 0.8.10, Copyright © 2000-2011 the FFmpeg developers
built on Oct 24 2025 12:36:04 with gcc 4.9.2
configuration: --prefix=/root/daily_build/52x_10/5.2.x/LinkFS/usr --enable-static --disable-shared --disable-yasm --enable-gpl --disable-decoder=snow --disable-encoder=snow --disable-decoder=ac3 --disable-decoder=ac3_fixed --disable-decoder=eac3 --disable-decoder=dca --disable-decoder=truehd --disable-encoder=ac3 --disable-encoder=ac3_fixed --disable-encoder=eac3 --disable-encoder=dca --disable-decoder=hevc --disable-decoder=hevc_cuvid --disable-encoder=hevc_nvenc --disable-encoder=nvenc_hevc --disable-decoder=h264 --disable-decoder=h264_cuvid --disable-encoder=libx264 --disable-encoder=libx264rgb --disable-encoder=h264_nvenc --disable-encoder=nvenc --disable-encoder=nvenc_h264 --disable-decoder=mpeg2video --disable-decoder=mpegvideo --disable-decoder=mpeg2_cuvid --disable-encoder=mpeg2video --disable-decoder=mpeg4 --disable-decoder=mpeg4_cuvid --disable-decoder=msmpeg4v1 --disable-decoder=msmpeg4v2 --disable-decoder=msmpeg4v3 --disable-encoder=mpeg4 --disable-encoder=msmpeg4v2 --disable-encoder=msmpeg4v3 --disable-decoder=mvc1 --disable-decoder=vc1 --disable-decoder=vc1_cuvid --disable-decoder=vc1image --disable-decoder=aac --disable-decoder=aac_fixed --disable-decoder=aac_latm --disable-encoder=aac --disable-decoder=on2avc --disable-encoder=ssa --disable-encoder=ass --disable-encoder=dvbsub --disable-encoder=dvdsub --disable-encoder=movtext --disable-encoder=srt --disable-encoder=subrip --disable-encoder=text --disable-encoder=webvtt --disable-encoder=xsub --disable-encoder=movtext --disable-decoder=ssa --disable-decoder=ass --disable-decoder=dvbsub --disable-decoder=dvdsub --disable-decoder=ccaption --disable-decoder=pgssub --disable-decoder=jacosub --disable-decoder=microdvd --disable-decoder=movtext --disable-decoder=mpl2 --disable-decoder=pjs --disable-decoder=realtext --disable-decoder=sami --disable-decoder=stl --disable-decoder=srt --disable-decoder=subrip --disable-decoder=subviewe --disable-decoder=subviewe --disable-decoder=text --disable-decoder=vplayer --disable-decoder=webvtt --disable-decoder=xsub --disable-decoder=ccaption --disable-decoder=movtext --disable-decoder=subviewer --disable-decoder=subviewer1 --disable-network --enable-small

-------------------------------------------------------------------------

I removed of cause the -ar 192 parameter.
Finally I compared the MinimServer "flac:wav24;192" against the Linn Akurate (fully upgraded bij Dr. Volk) internal "Digital Output Mode" set to 176/192kHz.
For now the Linn is my preference with the advantage that all Linn inputs will be upsampled ending up in the beautiful Topping D900 DAC and Topping Ext90 Pre-Amp.

[simoncn]

Thanks for confirming this (libsoxr).

It isn't surprising that a high-quality hardware upsampler running on a Linn is better than upsampling performed by free software running on a NAS.

[z624391]

But the MinimServer ffmpeg solution is not that bad. Keep in mind that for me MinimServer overclassed the Roon App!

And with all the information on the internet I learned a lot about the streaming proces (although you must put all the pieces together to get a complete story).

Thx for your guiding Simon.