transcoder order

[bpa]

> I want to check (I feel it is probably OK) are there any situation
> where PR might undo the effect of the custom conf.

Philippe suggested making this an option (I've enabled it by default).
Users who want to override default behaviour could turn this off.

Option is good for developers to allow reversion.

However it is a setting for a naive user to twiddle and have unexpected and hard to debug results.
Could there be situation where two plugins needs the setting set differently and so require one plugin to be updated.
I just need to think through behaviour in possible different scenarios

[philippe_44]

Option is good for developers to allow reversion.

However it is a setting for a naive user to twiddle and have unexpected and hard to debug results.
Could there be situation where two plugins needs the setting set differently and so require one plugin to be updated.
I just need to think through behaviour in possible different scenarios

I’m wondering if we are not talking about 2 different settings/options.

The one to evaluate the codecs matching with the native option first would be user -accessible, and disabled by default so that we keep the existing behavior. I would enable it as I prefer audio passthrough and even more direct player connection when possible, but that is a personal choice.

The one where a plugin can force transcode for certain formats, is not an option in Tidal today, it’s hard coded for flac when sending to ip3k players but that’s what you have in 8.0.1 (except that is applies for all players). I’m not sure it should be a user-level decision, but I leave that to you Michael

[mherger]

> The one to evaluate the codecs matching with the native option first
> would be user -accessible, and disabled by default so that we keep the
> existing behavior. I would enable it as I prefer audio passthrough and
> even more direct player connection when possible, but that is a personal
> choice.

I actually enabled it by default... (did you see my commit?)

> The one where a plugin can force transcode for certain formats, is not
> an option in Tidal today, it’s hard coded for flac when sending to ip3k
> players but that’s what you have in 8.0.1 (except that is applies for
> all players). I’m not sure it should be a user-level decision, but I
> leave that to you Michael

No, that's on all the time. No setting. Because the experience when
there's a hiccup sucks. Nobody wants to go search for that pref. It
should just work.

--

Michael

[philippe_44]

> The one to evaluate the codecs matching with the native option first
> would be user -accessible, and disabled by default so that we keep the
> existing behavior. I would enable it as I prefer audio passthrough and
> even more direct player connection when possible, but that is a personal
> choice.

I actually enabled it by default... (did you see my commit?)

Now I did - all good. It works fine on my devices BTW

> The one where a plugin can force transcode for certain formats, is not
> an option in Tidal today, it’s hard coded for flac when sending to ip3k
> players but that’s what you have in 8.0.1 (except that is applies for
> all players). I’m not sure it should be a user-level decision, but I
> leave that to you Michael

No, that's on all the time. No setting. Because the experience when
there's a hiccup sucks. Nobody wants to go search for that pref. It
should just work.

--

Michael

Agreed

[philippe_44]

One thing I'm wondering: with the news "native" rule, on ip3k players, when playing aif, it goes through but wav do not as the ip3k do not report a "wav" codec, so we default to "wav flc" rule. Shouldn't we add one exception that if source format is "wav", then we should also promote "pcm" codec at the top?

[Ron F.]

Running slimserver 8.2 on my home server, squeezelite client on an RPi Zero W, release v1.9.8-1307.

Streaming Spotify, 320kbps Ogg Vorbis. If I start squeezelite normally, is see that native support for Ogg is used. If however, I start squeezelite with the option "-c flac,ogg,pcm" then what I see is quite different, using More Info: File Format: Ogg Vorbis (Spotify) (Converted to FLAC). The Now Playing display in material reports that the stream is 705kbps CBR, Ogg Vorbis.

I checked File Types, (using the material skin,) and the option "Prefer native format" is checked and the transcode rule for Ogg -> Ogg is (I,F) Native, so everything looks correct, but the server is transcoding to FLAC anyway.

If I switch the squeezelite codec order using "-c ogg,flac,pcm" then LMS streams Ogg instead of transcoding to FLAC.

I thought "native" should take precedence over the client's codec order if the option to prefer native decoding was selected? I am confused however, and I won't pretend to understand how this is supposed to work.

[slartibartfast]

Running slimserver 8.2 on my home server, squeezelite client on an RPi Zero W, release v1.9.8-1307.

Streaming Spotify, 320kbps Ogg Vorbis. If I start squeezelite normally, is see that native support for Ogg is used. If however, I start squeezelite with the option "-c flac,ogg,pcm" then what I see is quite different, using More Info: File Format: Ogg Vorbis (Spotify) (Converted to FLAC). The Now Playing display in material reports that the stream is 705kbps CBR, Ogg Vorbis.

I checked File Types, (using the material skin,) and the option "Prefer native format" is checked and the transcode rule for Ogg -> Ogg is (I,F) Native, so everything looks correct, but the server is transcoding to FLAC anyway.

If I switch the squeezelite codec order using "-c ogg,flac,pcm" then LMS streams Ogg instead of transcoding to FLAC.

I thought "native" should take precedence over the client's codec order if the option to prefer native decoding was selected? I am confused however, and I won't pretend to understand how this is supposed to work.

Just a thought. Are you looking at the right OGG? The one you want is under Spotty in the file types settings.

Sent from my Pixel 3a using Tapatalk

[philippe_44]

Running slimserver 8.2 on my home server, squeezelite client on an RPi Zero W, release v1.9.8-1307.

Streaming Spotify, 320kbps Ogg Vorbis. If I start squeezelite normally, is see that native support for Ogg is used. If however, I start squeezelite with the option "-c flac,ogg,pcm" then what I see is quite different, using More Info: File Format: Ogg Vorbis (Spotify) (Converted to FLAC). The Now Playing display in material reports that the stream is 705kbps CBR, Ogg Vorbis.

I checked File Types, (using the material skin,) and the option "Prefer native format" is checked and the transcode rule for Ogg -> Ogg is (I,F) Native, so everything looks correct, but the server is transcoding to FLAC anyway.

If I switch the squeezelite codec order using "-c ogg,flac,pcm" then LMS streams Ogg instead of transcoding to FLAC.

I thought "native" should take precedence over the client's codec order if the option to prefer native decoding was selected? I am confused however, and I won't pretend to understand how this is supposed to work.

You observation is correct but the reason is that Spotty uses a custom-convert.conf file where its "native" format is 'spt'. The function that looks for a transcoder use as input the native format and all formats supported by the player. It re-orders the search to start with the native one (when option set). As 'spt' is never a supported format, then the re-ordering does not do anything useful.

[philippe_44]

But I had the idea of a simple patch that solves that issues - I'll submit that to the Master

[mherger]

> Streaming Spotify, 320kbps Ogg Vorbis. If I start squeezelite normally,
> is see that native support for Ogg is used. If however, I start
> squeezelite with the option "-c flac,ogg,pcm" then what I see is quite
> different, using More Info: File Format: Ogg Vorbis (Spotify) (Converted
> to FLAC). The Now Playing display in material reports that the stream is
> 705kbps CBR, Ogg Vorbis.

As Philippe has outlined this is due to Spotty using a custom file type.
That said: native Ogg playback unfortunately will be removed again. The
change Philippe sponsored hasn't been accepted upstreams, and
maintaining it in my code is beyond me.

I'll look into Philippe's pull request anyway.

--

Michael