[Log Error] [Dissonance:Playback] (17:10:41.949) EncodedAudioBuffer: Voice Error: Encoded audio heap is getting very large (40 items)! Error ID: 59EE0102-FF75-467A-A50D-00BF670E9B8C

cliv3dev commented 4 years ago

Context

When a host (on Editor) and a client (on Oculus Quest) are connected together through WAN connxexion (WiFi->4G shared connexion by an iPhone for the Client and Ethernet->Optic fiber for the Host), I can get those errors on the Editor log as well as the Quest log. There was no discussion at the time, only background noise.

Expected Behavior

Should not have those logs

Actual Behavior

during connexion of less than 1 minute, this error was logged 3 times.

Steps to Reproduce

Provide a detailed set of steps to reproduce the problem

start a host on Win Editor with Unity 2019.3.11f1 and Mirror 12.12.0 and Thelepathy transport
start a client on Oculus Quest with same Unity version player and mirror and data
wait for connexion
let the 2 devices run for some time

Your Environment

Host: Windows 10, Unity 2019.3.11f1, i9, 16Gb RAM, GTX1960, ethernet connexion to Optic fiber box Client: Oculus Quest with v16 OS connected to iphone 8 with shared connexion to 4G network

Dissonance version used: e.g. v1.0.0 (check Assets/Dissonance/changelog) v6.4.6
Unity version: e.g. 5.3.5f1 Personal (check Help > About Unity) Unity 2019.3.11f1,
Editor Operating System and version: e.g. Microsoft Windows 10 Pro 10.0.10586 (x64) Win10 Pro, 18362.778 (x64)
Build Settings: Tell us the build settings/platforms you have encountered this issue with e.g. Windows, x86_64, Development Build Oculus Quest/Android ARM64, IL2CPP, Master, no Dev build, LZ4HC Quest Realcast-Avatars Log.txt
Link to your project: If your project (or a bug reproduction project) is available, link it here

cliv3dev commented 4 years ago

WinEditor(Host) Realcast-Avatars Log.txt

martindevans commented 4 years ago

The log shows this message: EventQueue: Large number of received packets pending dispatch (26) which indicates that 26 packets of data arrived from the network within the same frame! This will cause the Encoded audio heap is getting very large error. There could be some kind of network error, or a large delay (e.g. extremely low frame rate) on the sending side causing all these packets to buffer up. This could be caused by the Telepathy backend that you're using - telepathy isn't really supported for voice because it's TCP but voice needs unreliable packets (i.e. UDP). Using a TCP connection can make small network errors much worse (e.g. a small interruption causes a large backlog in packets).

Could you try upgrading to Dissonance 7.0.1 (released yesterday) and running the demo scene with the Ignorance Mirror backend. If the problem persists please increase the Dissonance log level to Debug for all categories and send me another log.

cliv3dev commented 4 years ago

I can't use Ignorance, since they don't seem to support HoloLens2 platform (UWP). But I can test with LiteLNetLib4Mirror which has an unreliable UDP channel...

martindevans commented 4 years ago

LiteNetLib4Mirror should work just as well as ignorance :)

cliv3dev commented 4 years ago

Switched to LiteNetLib4mirror and testing localy on the same PC : Editor/Host and a standalone/Client on local network. I have those errors: [Dissonance:Playback] (20:10:14.125) EncodedAudioBuffer: Error: Received a very late packet (41 packets late)! This is probably a bug in Dissonance, we're sorry! Please report the bug on the issue tracker "https://github.com/Placeholder-Software/Dissonance/issues". You could also seek help on the community at "http://placeholder-software.co.uk/dissonance/community" to get help for a temporary workaround. Error ID: 30EF1B03-7BBC-49D3-A23E-6E84781FF29F Had this warning in the log 6 times in a row.

martindevans commented 4 years ago

Can you reproduce this error in the Mirror demo scene? usually an overflowing buffer like this indicates one of three things:

Performance issues on the sending side - low frame rate caused a lot of audio to be sent at once, overflowing the receive buffer.
Network level issue - lots of packets were buffered up in the network for some reason (e.g. congestion)
Performance issues on the receiving side - e.g. low frame rate causes audio playback to fall behind.

In your test (local network to same PC) I think we can safely eliminate a network level error. So it could be a performance issue in your scene (if look look in the profiler don't look at overall average frame rate, instead look at the worst frame times). Also check that you have Run In background enabled!

martindevans commented 4 years ago

Hi cliv3dev, did you make any progress reproducing this error in a simpler scene?

cliv3dev commented 4 years ago

Sorry Martin, didn’t take the time to reproduce it…

Cheers,

De : Martin Evans notifications@github.com Envoyé : Thursday, June 4, 2020 5:11:25 PM À : Placeholder-Software/Dissonance Dissonance@noreply.github.com Cc : Diego Fernandez-Bravo diego@realcast.io; Author author@noreply.github.com Objet : Re: [Placeholder-Software/Dissonance] [Log Error] [Dissonance:Playback] (17:10:41.949) EncodedAudioBuffer: Voice Error: Encoded audio heap is getting very large (40 items)! Error ID: 59EE0102-FF75-467A-A50D-00BF670E9B8C (#192)

Hi cliv3dev, did you make any progress reproducing this error in a simpler scene?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/Placeholder-Software/Dissonance/issues/192#issuecomment-638914194, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AFIM27KZENX4EG6AWWGWNPLRU62R3ANCNFSM4M2UEZRQ.

Placeholder-Software / Dissonance