Sonic Analysis on RPI4

Hello, I’m running a plex server on RPI4 (ARM device).
This RPI is also a private cloud serveur running nextcould.

Yesterday I bought a Lifetime plex pass because I was interested in the plexamp app and especially the Sonic analysis feature.

Today I discover that ARM devices are not supported this feature. I also read much topics about this on this forum.

@elan , I would like to know If it’s planned to enable sonic analysis feature on Rpi devices (even if it’s very slow)

sorry, no current plans, it’s just too slow.

Not sure to understand. Does it mean that it will be never implemented or just not planned yet?

We’re not planning on implementing it because the result would be way to slow.

This is a huge disappointment.

With regards speed, I have Windows machines slower than the RPi4 and I don’t really care how many days it takes to complete a full analysis so speed

Couple that with the astronomical increase in electricity prices here in Europe, people will not want to/be able to afford to run a huge Windows PC for this due to costs.

Many will be turning to RPi4 and other Arm systems to save running costs if no transcoding is required.

Tha fact you mentioned Rosetta 2 reveals the reason is that the libraries are written for x86 chips, not ARM, so speed may not be the actual issue here.

Whatever it is, it’s a problem that is likely to grow in the coming months as people try to balance their electricity bills with everything else.

1 Like

Hello,

I confirm that I’m disappointed too. :frowning:

Xyleon is right, I won’t be running a PC since I choosed to run a RPI4 because of it’s low power consumption.

I really think that ARM device are the future.

I don’t really care if this analysis take many months or years.

I’ll add that I too am disappointed. As @bpollet wrote, I don’t really care if the analysis takes a long time.

I do appreciate the candor though and the other good features of plex/plexamp.

Hello, I would like to bring different possible approaches to this problem:

  • Compile the needed libraries to ARM when compiling the Plex Server. This would require the source code of those libraries available as well as being able to compile into ARM. Not really sure about this one.
  • Use another libraries as a replacement of the unavailable ones. For most people it doesn’t matter if they are slower or less optimized as long as they, at least, work. Maybe there’s no drop-in replacement or a replacement at all, but I think it’s worth the effort to at least try.
  • Run the sonic analysis as a different external service. This could be a Docker container, or a remote x86 machine that runs a small API and the binaries required to perform the analysis. The Plex server sends the audio file and the API responds with the result of the analysis. This would be a similar approach as what Grafana does with their renderer plugin to generate images of graphs. The only problem I see is the added (but optional) complexity of setting up a different server to perform this task and the bandwidth usage.

To my point of view option n°2 should be the best one.

I’ll just add my wish to have sonic analysis on RPi4 - locally however slow, or a way to ship the data over from another server, or use an external analyzer - perhaps on another Plex media server; I think this last option would make most sense.

My main server is a powerful enough x64 NAS (PMS in an Oracle Linux 8 vm, media read-only NFS mounted), but I also have a Raspberry Pi 4B (8GB) with PMS and headless Plexamp - with an 1TB USB SSD holding a copy of our FLAC library and a small USB DAC/DDC, it travels anywhere. Lossless streaming, controlled via Android Plexamp, with (near-)zero internet traffic. :slight_smile:

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.