Opened 7 years ago
Last modified 4 years ago
#17225 assigned defect
Audio distortion/skipping/stutter in Windows 7 Guest
Reported by: | LittleJeff1970 | Owned by: | pentagonik |
---|---|---|---|
Component: | audio | Version: | VirtualBox 5.2.0 |
Keywords: | distortion, stutter, skipping | Cc: | |
Guest type: | Windows | Host type: | Windows |
Description
Both my host and guest are Windows 7 Professional. If I use the HD Audio driver for the guest, the audio lags behind the video a few frames for any streaming video such as news or YouTube. Also, there is distortion, popping, crackling, skipping, etc. of the audio. After switching to AC97 ICH in the guest, the LAGGING problem goes away. However, the distortion persists.
I followed the steps for Audio Debugging in the Wiki. The WAV file also exhibits some distortion when played on the host, though perhaps not quite as noticeable as the distortion in the guest. I am attaching the VBox.log file.
Finally, it may be worth noting that NONE of these issues existed for me when I was using VBox version 5.0.26. Every version of 5.1 I have tried as well as the newest 5.2 has this problem.
Attachments (38)
Change History (119)
by , 7 years ago
comment:1 by , 7 years ago
Owner: | set to |
---|---|
Status: | new → assigned |
comment:3 by , 7 years ago
I have the same problem on my Windows 10 64 bit host (1703). The WAVs are flawless, but the audio distortion comes and goes, in a repeating pattern, every few minutes, while playing back in the guest (I have tested Windows 7 and Windows 10 64 bit). The way I reproduce it, I let Youtube play in the guest's browser (Edge/Firefox). Here are some logs from a debug OSE build: link, where the problem starts to show maybe 1 second before the 14:14:07.700000 time mark, when I took notice. Even when it sounds OK, it still hints to some almost inaudible audio distortion.
comment:4 by , 7 years ago
Confirm. I have Windows 10 host, and Linux Min 18 guest and also have sound hiccups/glitches/crackles on Mint inside VirtualBox 5.2.0-118431. What i try to do to fix the problem:
- change Audio device from AC97 to Intel HDA and back;
- changing some ALSA parameters;
- drop number of guest CPU;
- disable APIC, and nothing helps. But. Installing VirtualBox version 5.1.30-118389 resolves the problem. Hope this will be fixed in newer version of VirtualBox.
comment:5 by , 7 years ago
The same problem on Win 2012 hosts, Ticket #17309 And yes, 5.1.30 worked well.
comment:6 by , 7 years ago
Problem persists in 5.2.2. Guest Extensions do not affect the issue in any way.
comment:7 by , 7 years ago
I've just uploaded a new 5.2 test build which contains more fixes regarding "crackling" and audio distortions. Could you please give these builds a try and report back if these fix the issue for you? You can get the latest 5.2 test builds here: https://www.virtualbox.org/wiki/Testbuilds
Thank you!
comment:8 by , 7 years ago
I already tested and reported this sound problem in the forum here and here in the past. Thank you for releasing a new version (19405), so I could test this once again. I also repeated the test with the newer guest additions form 5.1.31 (19407), but this didn’t change anything.
Comparing to the old state the situation now is little bit better, but not OK. On my Win 8 host / Win 7 guest combo I still have distortion on small test sounds. The stream playback now has growing distortion for 10 seconds not every 90 seconds, but just at the beginning and then after four and a half minutes. So probably the buffer is bigger now?
On my Win 10 guest and host the distortion in the small sounds now came not immediately but after 5 or 10 seconds. I got the distortion at the beginning of the stream playback, but not in the first 10 minutes of playback. Probably I need to test longer. And there were no problems with pause this time, as I described it in forum. During all the tests there were no complete disappearance of the sound, where I have to switch it off and on again in the VM toolbar.
I would tell this is probably the right direction but not the result I hoped for. If you need some more information or some logs, don’t hesitate to tell me.
comment:9 by , 7 years ago
Thanks for the (fast) feedback! The Guest Additions are totally unrelated in that regard, no need to upgrade those.
Could you please supply the verbose VBox.log file and the .WAV files as shown here: https://www.virtualbox.org/wiki/AudioDebug
Thank you!
comment:10 by , 7 years ago
Thank you for your replay. Here are the results of my tests today, as you asked. First of all, dumped MP3 are all alright, just as I already told in the forum. So, I have to use Audacity on the host to record the real audio output there. Then one can hear the distortion. Here I put only the parts which demonstrates the problems.
On my Windows 7 guest I have had to wait for 4:30 minutes to hear the distortion in the stream. On Windows 10 it always happens at the beginning of the playback and then at first approach at 5 minutes and at second and third approach at 6:35. Small sounds are always distorted. The logs are also here.
If you need complete recording, or not compressed sound, I saved the original Audacity projects. Just tell me, where to upload them, because they are big. If you have some further questions – just ask.
by , 7 years ago
Attachment: | W10 stream test1.mp3 added |
---|
Windows 7 Audio stream test, distortion at the beginning
by , 7 years ago
Attachment: | W10 stream test2_1.mp3 added |
---|
Windows 7 Audio stream test, distortion at 6:35, part 1
by , 7 years ago
Attachment: | W10 stream test2_2.mp3 added |
---|
Windows 7 Audio stream test, distortion at 6:35, part 2
comment:11 by , 7 years ago
Here's my log: link. When I first paused the VM, the audio crackling was just starting to manifest. The 2nd time I paused the VM, the audio crackling was already manifesting itself for several seconds. After each unpause, the VM resumes normal audio playback. Here's the WAV file: link, but it's flawless, I can't hear anything bad when I play it.
Here's how it's being heard on the host: link (recorded on a different VM session, when playing the same audio track in the guest). You can also hear normal playback after VM pausing and resuming.
comment:12 by , 7 years ago
I uploaded my WAV files and the complete recording from the host as an MP3. Here the one for Win7 and here for Win10.
A little more testing shows, that this crackling behavior is definitely usage dependent: in case of 64 KB/s stream playback it comes once in a while, it repeats now every 4 or 6 minutes. If you play some good quality MP3 or YouTube video in VM, the distortion is present nearly constantly, it’s never go away.
comment:13 by , 7 years ago
log+wav from Windows 2012 host/guest: Audio distorted in VM and in recording
comment:14 by , 7 years ago
I've just tried builds 5.2.3.119620 5.2.3.119658
The sound now is distorted even when clicking Playback devices -> Test.
Host: Win 2012R2 Guest OSs: Win2012R2, Win 7 Home
comment:16 by , 7 years ago
penntagonik, any progress on this?
Maybe you can highlight what you think may be causing the issue and possible problematic src code so that I (and possibly others) can also take a look to help identify the problem ?
Regards
comment:17 by , 7 years ago
sha384
You should be trying the latest test builds. Also, if you want to see the source code that's being worked on, you should check the Timeline, for the audio related changesets.
comment:18 by , 7 years ago
Thanks socratis.
I have been running some debug tools in a Win7 VM, here are a few observations:
- CPU-Z revealed a clock that is lower than the actual CPU clock. The VM CPU clocks are showing 3000 MHz clocks w/ a x3 multiplier and a bus speed of 1000 MHz. The actual CPU clock is 4299MHz, with a x43 multiplier and a bus speed of 100Mhz. Power settings are in high performance mode. Execution cap was at 100% if it matters.
I am running a sandy bridge i5-2500k for your information.
- There are some pretty bad interrupt latency issues, which may explain why the audio is popping and crackling. I haven't honed into the cause of it, but it looks like it may be a driver DPC/interrupt that is taking too long to process, and thereby slowing down other routines, including audio which is time sensitive.
The base test had all auxiliary features disabled (usb, serial, optical, networking, drag and drop, clipboard sharing and so on).
DPCs/ISRs would take in the region of 300-500 micro-seconds (idle), with frequent spikes to 5ms (idle), and up to 40-50ms (w/ 200ms peaks!) when under load.
When the audio input was turned off the DPCs/ISRs would range between 100-350 microseconds (idle), with spikes at 500 microseconds (idle), and peaks of 5-10ms when under load.
So it looks like the audio interrupts may need some tweaking, and the CPU clock-rates corrected - 1200MHz is a lot of cycles going down the drain. With that said, some general refractoring and code optimisation may also be due across the board.
comment:19 by , 7 years ago
I added some test result files with logs in the forum. If somebody is interested, that one can start to hear some small crackling at the beginning of the small files playback in the VM.
follow-up: 21 comment:20 by , 7 years ago
I've just tried build 5.2.7-120325 The sound is distorted even when clicking Playback devices -> Test. Host: Win 2012R2 Guest OSs: Win2012R2, Win 7 Home
And also I've noticed that in recent versions the WAV file does not appear anywhere after setting extra audio logging.
It would be helpful if you at least notify us here when we should check the new build. Thanks.
follow-up: 22 comment:21 by , 7 years ago
Replying to denZXabc:
It would be helpful if you at least notify us here when we should check the new build.
Not all test builds are audio-related. Or targeted to your specific problem. They might contain fixes that are not meant for your particular issue. It's not a "Hey everyone, new test build" every time. When such a test build requires your attention/feedback, an announcement is usually made, either personally, or as a general announcement.
comment:22 by , 7 years ago
Yes socratis, that's exactly what I mean. I just haven't seen notifications here, besides the very first one. Thanks for reply. Replying to socratis:
Replying to denZXabc:
It would be helpful if you at least notify us here when we should check the new build.
Not all test builds are audio-related. Or targeted to your specific problem. They might contain fixes that are not meant for your particular issue. It's not a "Hey everyone, new test build" every time. When such a test build requires your attention/feedback, an announcement is usually made, either personally, or as a general announcement.
by , 7 years ago
Attachment: | VBox.2.zip added |
---|
Host: Fedora 27, Vbox 5.2.6, Guest: Windows 8, Affected soft: Skype for business
comment:23 by , 7 years ago
Audio issue with vbox 5.2.6 and win 8 + skypeforbusiness. After a call to Skype, there is absolutely no sound. Log file for 5.2.6 in attachment.
There is a same issue with vbox 5.2.4 too, but here is a little different behavior: The sound disappears not immediately but after a while.
comment:25 by , 7 years ago
No, not really. I tested with current 120684 test build and with some 1205xx previously, at my side the situation is the same. There are some development in the Subversion repository, but right now there are no final achievement in this case, that sound issue is the same.
comment:26 by , 7 years ago
Just tested freshly released build 120822, it works fine on my Windows 10 host and guest combo. On my Windows 8 host and Windows 7 guest system I can hear something like a sound of an air bubble popping from the water. This sound arrives randomly nearly every 10 seconds. And this is everything I can report about as an error. Comparing to the previous state of the sound in the 5.2.x trunk this is nearly perfect. For sure this is right direction of the development.
comment:28 by , 7 years ago
@others Please have a try with the latest 5.2 test build 120822, which is located here: Testbuilds. Thank you!
comment:29 by , 7 years ago
I'm reopening this ticket with the following comments for the 120822 test build:
- No audio crackling so far, that's very good.
- I can't play back the wav files that VirtualBox outputs when audio debugging is enabled. That's a new issue. Winamp hangs and Windows Media Player 12 refuses to open the files. VLC seems to be the only player that has no issue with it.
- VirtualBox saves the wav files to C:\TEMP. How does it choose this location? Because %TEMP% and %TMP% are not pointing to C:\TEMP on my system.
- Audio stops after minutes of playback (actual cases, after 11 minutes and 29 seconds, after 14 minutes and 38 seconds). It seems to occur every time. If I stop and restart the playback, it works sometimes, but if it doesn't, I have to try until the playback starts. I've never encountered such a problem.
Audio dump files and the VM log: link
comment:30 by , 7 years ago
I‘m sorry, but I have to tell that I was a little bit too optimistic about new build. I also have same kind of problems. There are no more distortion here, but the sound simply disappears. On my Windows 10 host and guest combination it happens after 11 or 12 minutes, on my Windows 7 guest on Windows 8 host even earlier, after 2 or 3 minutes. The playback in the VM goes on, but there are no sound outside. I have to stop or pause the playback and start it again to have the sound. It is better than before, but still not so good.
If one take a look in the logs one will see, that sound stream became asynchronous and was stopped after a while. That is what caused the disappearance of the sound. I publish booth of my logs here.
comment:31 by , 7 years ago
@MihaiH I've added instructions to the audio debug Wiki page about how to tweak the debug output directory accordingly: https://www.virtualbox.org/wiki/AudioDebug.
comment:32 by , 7 years ago
@boxer01 Thanks for the dumps -- I've checked the files and only could hear some white noise playing back. Was this intentional, e.g. did you play back some sort of test file? Both, the device emulation and the backend output, seem to stop after about 11 minutes, so not quite sure yet what has caused this.
comment:33 by , 7 years ago
I‘m not quite sure what you are talking about. If you mean sound MP3 dumps from 2 month ago (beginning of December) after comment number ten here and something I posted in the forum, then it isn’t a white noise, I just checked. I first test with small sounds like sound when one changes sound volume with mouse over the speaker icon in the tray and test sound on playback devices in control panel and then sounds. Here you can hear distortion nearly immediately. The second part of the test is the Internet audio stream playback where you need some time for distortion to come. I wrote all this test parkour in one of the forum’s post I linked from here. And in these recordings directly from host sound card with Audacity one can clearly hear the distortion.
But this is not important after the version 120822, because here the situation has changed. We have no more distortion, now the sound disappears completely in spite of the fact that in the VM the program continue to play. It’s like a mute for complete VM. Only if I stop or pause the playback in the program and then start it again, the sound appears. And if you take a look at the logs I posted yesterday (mid-February) you will immediately see the cause of this. The buffer became too full or too asynchronous and at some point VM audio back-end decides to stop playback. Search for "stoping" here. So it’s still the buffer problem, but now it shows up as an audio stop and not as some kind of distortion.
If you need, I can post the MP3 of current dumps. But here everything is fine as usual, The recording stops at the very same moment when VM playback stops and continues when I restarted it. BTW the 120865 build has the same behavior.
comment:34 by , 7 years ago
Hello Version 5.2.97 r120712 (Qt5.6.2) https://www.virtualbox.org/wiki/Testbuilds
The sound is correct. No distortion. No mute. The test lasted about 30 - 40 minutes.
Host: Windows 10.0.16299.248 Guest: Ubuntu 16.04.4 LTS
I'm continuing testing.
comment:35 by , 7 years ago
Using the new test build (120865), I get small audio hiccups and the audio stop issue. Link for the dumped WAV files. With this particular VM session (check the log), I've paused the VM a few seconds after the audio stopped, then resumed it and audio playback restarted with a distortion at the start.
by , 7 years ago
comment:36 by , 7 years ago
VirtualBox 5.2.8 is affected by similar audio crackling defects as in comment #11.
by , 7 years ago
Attachment: | VBox.2.log added |
---|
comment:37 by , 7 years ago
I‘ve tested the new 5.2.8 build. On my Windows 8 host / Windows 7 guest combination it works fine, but every change of load on the host or high CPU load is causing some hiccups with sound in the guest. But even after hour of running I can't’ hear any known problems like distortion or no sound at all. Maybe this is because of continuous issues caused by host’s CPU load.
On Windows 10 host and guest combination as somebody already wrote it’s now a combination of old distortion from 5.2.x line with sound disappearance somewhere after 10 minutes from last 5.2.7 builds. But now it’s not a disappearance of the sound but the distortion starts after this time. In the logs on both systems one can see that VB tries to catch up with the sound to no avail, sometimes time after time, sometime constantly.
I also tested the 121224 build. Here one can hear constant sound jumps just like sound from moving LP or CD player, when CD or LP can’t be stabilized. I have no logs from this version but I can create them if it’s needed. I also can upload the dumped sound files, even if there is nothing wrong to hear.
For 5.1.34 I created a separate ticket #17614
comment:38 by , 7 years ago
@boxer01 Can you please try to provide the audio debug files as noted here: https://www.virtualbox.org/wiki/AudioDebug.
Handling audio requires a lot of effort in a VM, as this is realtime stuff and therefore of course also is dependent on the host's current load. That is, the VM tries to catch up if it falls behind.
comment:39 by , 7 years ago
Here are the links: Windows 7
Win7_hdaDMAReadSD4-0.zipWin7_hdaStreamReadSD4-0.zip Win7_PlayNonInterleaved-0.zip Win7_StreamWrite-0.zip Win7_StreamWrite-1.zip
Windows 10
Win10_StreamWrite-1.7z Win10_hdaDMAReadSD4-0.7z Win10_hdaStreamReadSD4-0.7z Win10_StreamWrite-0.7z Win10_PlayNonInterleaved-0.7z
As usual, I make my offer to not just make a log and dump files but also record directly from the line out with Audacity so you can hear, what really happens and what kind of distortion is presented at the host in the end. Because dumped files are normally OK even if the sound output is distorted.
comment:40 by , 7 years ago
For a linux host: My win7-64 guest started having serious audio issues as well--from out of nowhere. I'm still not sure why it started, but I fixed it by killing pulse audio, moving all of the files in my /home/uname/.config/pulse, and restarting pulse audio. Now it's fine.
comment:41 by , 7 years ago
Well, it wasn't fine, but I’m wondering whether or not the audio problems I’ve seen in Windows 7 virtual machines is a consequence of some mishandling of the multithreading implemented by the Ryzen 7 CPU in my new machine. I have that idea because the problems didn’t turn up until I installed a new BIOS on the Ryzen machine, and have had no problems at all using an identical VM hosted by an older AMD machine with the Fx8350 CPU that does not multithread. I will turn off the multithreading in the BIOS and report back.
The answer was not multithreading, but a new BIOS (4.60) for my x370 Taichi motherboard. Reverting to 4.40 seems to have fixed the issues.
4-11-2018 update: the BIOS was the problem. All audio functions working perfectly with the older BIOS.
comment:42 by , 7 years ago
Just tested with the latest development build 121438. Sound has constant distortion like an old LP player with jumping LP. Sometimes it sounds like an old type recorder which has some problems with the motor so the speed of the playback is slightly slower than needed. Both host and guest have nearly no CPU load. I put my logs and sound files here if somebody needs them.
hdaDMAReadSD4-0.7z hdaStreamReadSD4-0.7z PlayNonInterleaved-0.7z PlayNonInterleaved-3.7z StreamWrite-0.7z
comment:43 by , 7 years ago
This applies to windows xp and 2k too, so safe to say it affects all windows guests...
follow-up: 45 comment:44 by , 6 years ago
Any new updates on this issue? I'm tracking this problem since half a year now. I also tried a lot of things for myself without any luck. The last version that works for me without audio distortions is 5.1.26. This is the version where I'm currently stuck to until the problem gets solved.
The solution for this problem could probably be found if someone would have a look at the changes made between 5.1.26 and 5.1.28.
comment:45 by , 6 years ago
Replying to Mirji:
The solution for this problem could probably be found if someone would have a look at the changes made between 5.1.26 and 5.1.28.
Hey that's an idea! I wonder why the developers didn't think of that one before...
follow-up: 49 comment:46 by , 6 years ago
I can understand both sides, the users and the developers, but I think that after more than a half year period things looks different now. Users are more than a little bit worried now, because the important feature of the product they are using is broken for a long time. And this behavior started already in the release candidates at the end of September 2017. Nerveless this version was released with not working audio. After some experiments we still has no working audio in this version. So it not working differently as at the beginning for over then 6 month, but is still not usable. Meanwhile the current 5.1 version is also delivered with the broken sound emulation.
I know that socratis is a little bit sarcastic about such kind of things like critics and good minded advises. It’s OK, but after more than half year I can’t find this sarcasm funny anymore. His constant argument in this case is to take a look at the repository and find out how many commits was done in the audio corner. This also don’t work anymore: in the last 7 or 8 weeks most of the changes were refactoring of the code and the most of them was in UI part. The last change for this issue was done seven weeks ago so every test version and the last release has the same behavior. So for me it looks like sound emulation isn’t important for this product. I don't know what are paying users using this for, but it looks like sound isn’t important for them. I can’t imagine some scenario, where some desktop OS virtualization for the user wouldn’t use some sound playback. But probably all the commercial users use this software only for some GUI less virtual installation of some servers on some blade servers.
Beside I take a look on the whole situation. This ticket has nearly no activity in last two month. Same for the topic in the forum. My question: what all of this people doing now? Do they change the product completely and went to some concurrent one? Like this angry DJ at the year’s beginning? Or do they hold on the last working 5.1 version? So here then another question, which should be obvious for everybody: what they would do, if MS would make some big changes in the OS, so VB wouldn’t work, partly or at all? If anything would be patched, then the supported versions, and both of them has no working audio back end. Some of the moderators are still using 4.3 in productive environment, but what would they do after the next Meltdown / Spectre path from MS would render this version unusable? Version 4.3 is long out of support. And soon after the 5.3 release same would happen for the 5.1. That are some questions I have for this topic.
One of the socratis advises is to buy commercial support and order the path then. I know that we are not paying money for this. But we spending time testing it and trying to make it better. And we are still loyal to this kind of game. I would like to use it further because I see some advantages over the products of Vmware of MS. But in the end, if nothing else would help, I’ll go back to another one. So the question, I also sometimes ask myself: is this another OSS product from the Sun, which Oracle is trying to kill, accidentally or not? Everybody knows that the most of OSS products after purchase either died or went to another community under the different name because Oracle had no interest in them. Is it time for Vbox now to go the same path? Or the future of the product is to be in the small corner of the server virtualization on Oracle hardware?
The question I already asked: why the sound emulation was separated in the different thread at the first place, what are the advantages? Because we are all here because of the disadvantages of this decision.
That are my questions after 5 sub-releases and 6 month of this ticket. But probably we wouldn’t know the answer, even socratis, so the time will tell as usual.
comment:47 by , 6 years ago
Another question: after the last changes I had a look at the logs and I saw that for every sound playback the emulation back end is re-initialized. And every time the 8K buffer is reserved for the sound. If I’m right then my question is: why only 8K? This is only half a second on 128 kbit/s stream and even less on some higher bit-rates. So no wonder that every change of CPU consumption leads to some buffer over- or under-runs. Shouldn’t we can reserve more or have some settings for this?
I’ll create some head-up tickets (#17737 and #17738) for the last released 5.1 and 5.2 version so this issue would be seen as not resolved.
follow-up: 53 comment:48 by , 6 years ago
I honestly don't understand your way of thinking boxer01. Honestly... And it's not because I'm not trying...
You created two additional duplicate tickets for what exactly?
- To draw attention to the fact that the audio is not working as it should? Do you believe that this an issue that's not being worked on? Proof is required that this is the case. Uneducated speculation, or your hunch is not proof.
- To ease the burden of the developers? Help them out? In which universe creating additional, unnecessary load (reading, figuring out, closing ticket as duplicate; twice) does that not distract the developers from what they have to work on?
- To create additional psychological pressure to the developers? You've been doing that just fine so far (and some), there's no need for additional tickets.
I wouldn't reply to your pleas to fix this (like the devs are ignoring it), if you hadn't referred to me by name; and that's the second time, in two separate tickets! Here's the deal: I'm trying to help you and the others with updates and what's the situation at the moment.
As I've said (actually the developers, that's where I get my info), the biggest remaining problem is the Windows hosts backend. Frankly, I couldn't care less, I'm on OSX. And I should stop informing you guys of the progress, maybe I'm making things worse when I say "Hey, new build, fixes audio issues", maybe I should just shut the hell up and leave you in the dark! I'm looking at the Timeline (have you ever?) to see what they're fighting with. And it's a big scary monster.
Do you think that the "buffer" suggestion of yours is what's going to make it "sing"? Or that you can actually help? Go for it!!! Get the source code, implement the changes that you have in mind (obviously), test the changes (in all the platforms that are affected), submit the patches. Profit!
Instead of ... whatever you're doing right now. Whatever it is, it's not in the right direction, it's counter productive...
follow-up: 54 comment:49 by , 6 years ago
Replying to boxer01:
My question: what all of this people doing now?
Personally I use use 5.1.30. Sound works just fine there in Win7 guest on Win2012 host. When a new VirtualBox version comes out I try it, the problem is still there, I roll back. The ticket is alive, socratis writes updates now and then, so I had no reasons for panic and histerics. It's clear that the problem is nasty and needs time to be fixed properly. To calm you down the fun fact. The goddamn microsoft has finally fixed line-breaks problem in their Notepad after 30 years from inventing the problem. And you're complaining about 6 month.
But probably all the commercial users use this software only for some GUI less virtual installation of some servers on some blade servers.
I bet it's 95%+ use cases.
comment:50 by , 6 years ago
I've just tried versions 5.1.32, 5.1.34, 5.1.36, 5.1.38. They're all affected. Rolled back to 5.1.30 again. UPD: added 5.1.38
comment:51 by , 6 years ago
follow-up: 55 comment:52 by , 6 years ago
Just tested 124848 test build and this is the best 5.2 build since RC. My Windows 10 host and guest combination got no glitches since hours and on my Windows 7 guest on Windows 8 pair there were some small glitches after 4 or 5 minutes, but they were solved in less than a second. If somebody needs the description of the problems on the earlier builds (like 1242xx, 124553, 124628 or 124809) I can put it here. But because the current test build works fine, I think that there are probably no need for this.
Anyway I would like to thank a lot those people (probably pentagonik) who worked on this task for this afford. Don’t stop and put it into the next release. And please finish all the changes to make the buffer use the parameters from the machine’s file.
comment:53 by , 6 years ago
Replying to socratis:
I know that this was kind of rhetoric post, but I'll answer this anyway. As I wrote, I didn‘t see any big changes in SVN or in test or releases for weeks or for months. There also were no responses here or in forum. Because my para-psychological abilities are now on they historical low, I couldn’t know if these issues were ignored or simply pushed back because of lack of time. But I could imagine, that because there were no reports from the users on this one for very long time, that the developer team could think that everything is fine about the sound now. Once again, I can’t look into the heads of the developers and team managers, all information which I have is SVN, forum and result builds.
Because of this I decided to file another ticket for the newly released version. I thought: if I’m wrong (and time proved that I were wrong), then this is simply another duplicate and it would be closed as such in a minutes. Should I be right, then this is information which should reach the developers. So I don’t know if this could be used here, but I would be better safe than sorry. I saw many times the situation, where everybody thought that somebody else took care of the problem or at least informed about this, and at the end nobody does it.
My intention was to inform, not to build pressure or to distract somebody. If I or somebody else put some information or good minded advise here, then it was not for a distraction of somebody else. We are not the enemies, just the friends telling the not so fine truth. About your advice to do corrections myself or to take a look in the repository: I would if I could. It’s not enough to have knowledge in some programming language, one should also know the intern structure and development way of the project, and this would take a time. By the way: you know that the public accessible repository is only the read only mirror of the internal one? This is only explanation for the fact, that there could be no changes for three or four weeks and then there are hundreds of the change sets appears on one day.
I would like to thank you for informing us about any changes in this topic. But since there are such problems I already tested every test or development release available for the public to simply see if something was changed. On the current version of the Windows 10 one can only use version 5.1.24 of the VB or later; the previously released version are simply not compatible. But in the 5.1.32 the sound was already broken. And this is just a question of time when MS would make another questions and all the 5.1 versions wouldn’t run on the Windows 10.
As I already wrote, taking the look into the repository, there were no big changes for the audio between the march and august. But since last month there are many of them and I’m glad that somebody took care of this problem.
comment:54 by , 6 years ago
Replying to denZXabc:
I also use the 5.1.30 for non-test purposes because this was the last version with working sound. And there is no need to calm because I’m not stressed. I know that it’s easy to tell somebody that he or she should be happy to lost only one foot, because it could be both of them. But this shouldn’t be the reason to cut them both even if this isn’t required.
As of your notepad example, it’s really good. Because I simply don’t care about those changes, I’m using Notepad++ over the decade and should something goes wrong, there are at least two portable alternatives and even more for installation. The same thing is with paint and screen capture tools: there are many alternatives for this, so one should use them instead of delivered applications.
But for the VB I don’t know any alternative, changing to Hyper V or VMWare would be such a regress for me. The proper example here is the Firefox: during the transition to the new technology last year (Quantum) some things were broken by decision. So now, ten month after the release, I still miss some extensions like TMP with functionality to froze or protect the tab. It was so easily achieved in the original Firefox but now there are no possibility to protect the tab from refreshing or closing. Just because Mozilla decided not to give the API or that such development should have a low priority. I found a decent replacement to couple of my other extensions, but still miss the download manager like DTA. And because some API for this one are also missing, it would take at least another year to get this. And not, Chrome and other browsers are still light years away from the Firefox even after ten years of development, in the terms of extension. So there are also no decent replacement for this piece of software here. Let us hope that this issues with sound would be resolved at the end an everything would be fine.
comment:55 by , 6 years ago
Replying to boxer01:
Sorry to inform you, but I already found the first problem with this test release – the host audio output switch. In my case I switch from Speakers to the Bluetooth and back. The guest can‘t properly switch so the playback in the guest constantly plays the last couple of milliseconds. The only solution is to stop and start it again, sometimes twice.
After that the output appears in the new audio destination. But sometimes the buffer is not correctly initialized in this case and the crackling sound appears immediately. Now one need to stop and start the playback once or twice again and then everything is alright. If I switch back the whole thing should be repeated. Hope this can be easily reproduced and somehow corrected.
comment:56 by , 6 years ago
@boxer01 Thanks for all your (long) feedback -- appreciating it!
Regarding the "output switch" problem: Which host is this, and which backend has been used? Could you please provide a verbose log file of this as stated here? Thanks!
comment:57 by , 6 years ago
It’s Windows 10 1803 guest on same version host with HDA used as audio interface. Probably you can see this in the supplied logs.
I just tried three times to reproduce this with verbose logging, but I can‘t do this till the end. As soon as I switch the audio output by turning Bluetooth on or off, the VM simply hangs of. Nothing inside it show any reaction to the mouse or keyboard input, all you can do is minimize it. As soon as you are trying to save the state, the VM hangs of completely and only can be shut off by closing the window from taskbar or from the task manager in the host. In case of no verbose logging and logging audio recording I can go further by pausing and starting the playback in the guest some times, as I already wrote.
I put here all the logs and audio recordings which were produced, but I don’t know if this is complete files or not. I’m away from this test system for next 2 weeks, but you can write me back and I’ll do all the required test as soon as it would be possible. Thank you for your patience and effort in this case.
comment:58 by , 6 years ago
Thanks for the logs -- I see that you're using some "portable version" of VirtualBox. Where did you get this build from, how is this intended to being used? Generally we don't support such custom builds.
However, I already have an idea what's going on now, the problem is related to the so-called multimedia endpoint detection. I'll have a look at that.
comment:59 by , 6 years ago
I tried the release version VirtualBox-5.2.18-124319-Win. The sound out works perfect in Win 7 and Win 2012 guests on Win 2012 host. But the sound in (microphone) doesn't work at all. Should I file a new ticket or can it be handled as a part of this issue?
follow-up: 61 comment:60 by , 6 years ago
@denZXabc Did you also try the latest test build found here with regards to your microphone issue?
comment:61 by , 6 years ago
Thanks @pentagonik, In build "Version 5.2.97 r125396 (Qt5.6.2)" sound in/out works. Actually sounds seems work even better than in 5.1.30. Replying to pentagonik:
@denZXabc Did you also try the latest test build found here with regards to your microphone issue?
follow-up: 63 comment:62 by , 6 years ago
@denZXabc
The .97 builds are the development snapshots, not the test builds. The test build that 'pentagonik' was referring to must be 5.2.19 something...
comment:63 by , 6 years ago
In VirtualBox-5.2.19-125384-Win sound in/out works fine as well as in VirtualBox-5.2.97-125396-Win. I also tried sound enable/disable (VM menu Devices->Audio->Audio output): it works definitely better than in 5.1.x, but after several enable/disbale cycles the sound would crackle when it is disabled. Replying to socratis:
@denZXabc
The .97 builds are the development snapshots, not the test builds. The test build that 'pentagonik' was referring to must be 5.2.19 something...
follow-up: 65 comment:64 by , 6 years ago
Tried some other guest OSes.
Host OS: Windows Server 2012 R2 Host OS: Linux Mint 19 Cinnamon
VirtualBox Version 5.2.22 r126460 (Qt5.6.2)
Guest OS: Ubermix4 (Ubuntu 18.04), Sound: OK Guest OS: Win 7, Sound: OK Guest OS: Win 2012 R2, Sound: OK Guest OS: Win 10, Sound: broken Guest OS: Win 2019, Sound: broken
comment:65 by , 6 years ago
Replying to denZXabc:
Broken sound on Win 10/2019 guests probably related to Audio Channel.
comment:66 by , 6 years ago
VirtualBox 5.2.22 on Windows 7 host.
Audio is not in sync on Windows XP guest. There is always a 1 second delay. It is noticeable when start, pause or resume playing.
Disabling audio output sometimes causes crackles. It also slows down playback timer on Windows 7 guest and stops playback timer on Windows XP guest.
comment:67 by , 6 years ago
First of all thanks for the hard work.
As sound in Win10 like systems is broken i gave it a try on several versions of VirtualBox 5.2.22 r126460, 5.2.23 r127298, 6.0.0_BETA3 r127302: wav+logs
Interesting enough in 6.0.0 the sound is broken even with Windows2012 guest - a bit distorted and right channel gets mute sometimes.
follow-up: 70 comment:68 by , 6 years ago
I sorry to inform you, but the latest changes in the audio code put the distortion a little bit back. There are some changes about the new Windows 10 1809 default surround sound(@75606, @75962, @76049) and follow-ups (@75983, @76018, @76045, @76047, @76048, @76179, @76180); stream start optimization (@75980, @76157, @76188, @76189); random (@75964, @76019, @76022, @76158)
Here are more details: In the current 5.2.23 branch the distortion starts immediately at the beginning of the playback. It isn't so heavy as months ago and last for only 5 or 7 seconds. It repeats after something like 50 seconds. And this happens on both of my Windows 10 host and guest system and Windows 8 host and Windows 7 system. So the audio of 5.2 branch is currently once again broken.
The 6 branch test releases between the third beta and first release candidate were also broken, but now it’s fine. There is another problem here: back in September I already wrote about the problem by switching the host audio from speaker to Bluetooth and back. But back then one easily could simply stop or pause the playback and start it again. Then it went to the right audio device. In the current 5.2 and 6 test releases this one get worse, because now the playback stuck in this state simply forever. At least till you switch back, stop the playback, switch the device once again and start the playback. Only after this you can get the sound on the other device. I think that some initialization optimizations or caching causing this kind of behavior.
I created some logs and audio files for different version which I put here. One can clearly see the synchronization troubles in the logs, which causes the distortion, and the device switching is also here. As at the last time, trying to record some audio during the device switch hangs the VM completely; I only could go further by killing it by window closing or through the host task manager. Without audio debugging there is still this issue, but at least no hanging of the VM.
Windows 10 files:
VBox_5_Win10_1.7z VBox_5_Win10_2.7z VBox_5_Win10_3.7z Vbox_6_Win10_1.7z VBox_6_Win10_2.7z
Windows 7/8 files:
VBox_78_5.2.23-127298.7z
5.2.23-127309.7z
6.0.0_BETA3-127339_1.7z
6.0.0_BETA3-127339_2.7z
6.0.0_BETA3-127339_3.7z
6.0.0_RC1-127378_1.7z
6.0.0_RC1-127378_2.7z
6.0.0_RC1-127378_3.7z
by , 6 years ago
Attachment: | VBox_78_6.0.0_BETA3-127339_log.7z added |
---|
Win 7/8 6.0.0_BETA3-127339 logs
by , 6 years ago
Attachment: | VBox_5_log_Win10_2.7z added |
---|
Win10 VBox_5 logs; from BT back to speaker
by , 6 years ago
Attachment: | VBox_5_log_Win10_3.7z added |
---|
Win10 VBox 5 log, no switch, just distortion
by , 6 years ago
Attachment: | VBox_5_VBoxHardening_Win10_3.7z added |
---|
Win10 VBox 5 ahrdening log, no switch, just distortion
comment:69 by , 6 years ago
Upgrading from 5.2.22 to 5.2.24 has completely broken audio and video for me, Win7 host & Fedora 29 guest. I did notice audio was broken in the 5.2.23 test build but fixed it by backdating to 5.2.22.
5.2.24 seems to be much worse than the test build, but that is subjective. I am currently in a work environment and cannot allocate the time to look into it right now. Downgrading to 5.2.22 here has fixed it for me.
comment:70 by , 6 years ago
Replying to boxer01:
Just want to give the current state. No changes since my last output about the 5.2.23 and 6 betas, not for good. So playback in 6.0.2 (and .3) working good right now, there are no glitches or scratches here. Just a bit of it at the start of playback sometimes, but this is easily cured by stopping and starting the playback again. The 5.2.24 (and .25) playback is distorted at least once per minute for 5 or 7 seconds, just like in the previous beta versions. I put some audio files and logs from debug here.
My other issue is still here and even got worse since my last message: switching the device. Previously the playback just got stuck in the last second over and over again as one switched the host audio device during the playback. But it was solved by stopping and starting the playback once again. In the last tested betas 5 weeks ago the device was cached somehow, so one have to do more, like turning VM audio output off and on in the VM settings or waiting sometime. But now it’s stuck in the old device forever. No action could help here: stopping, pausing, starting once again, turning the audio output off and on, even saving the state of VM and start it again or switching back to the original host audio output device; the VM is just stuck in the old device and plays the last second forever. The only solution right now is to shutdown the VM (no restart) and start it again with new device as the host current active audio output device.
As usual the debug files are incomplete because as soon as I switch the devices in debug mode the VM hangs and could be only killed from the host’s task manager. I tried to test 4 scenarios: switch during playback and stopping the playback before the switch; I also tried to switch from speakers to the Bluetooth and in the either direction. But there is no difference between them all, neither in normal nor in debug mode, it just doesn't work the way it was in 5.1. By the way you don’t need to have something pesky like Bluetooth headphones or HDMI / Display-Port enabled display as an extra audio device: plugin the external headphones and switching to it from the internal speakers are enough to provoke the situation at least on my laptop.
Windows 10 files:
VBox_5.2.25-128236_Win10_1.7z VBox_5.2.25-128236_Win10_2.7z VBox_5.2.25-128236_Win10_3.7z VBox_5.2.25-128236_Win10_4.7z VBox_5.2.25-128236_Win10_5.7z
VBox_6.0.3-128259_Win10_1.7z VBox_6.0.3-128259_Win10_2.7z VBox_6.0.3-128259_Win10_4.7z
by , 6 years ago
Attachment: | VBox_5.2.25-128236_Win10_logs.7z added |
---|
5.2.25 logs, windows 10 systems
follow-up: 72 comment:71 by , 6 years ago
@boxer01 Thanks for your detailed feedback, really appreciating it! I'll have a look at all the provided files as soon as possible.
comment:72 by , 6 years ago
Replying to pentagonik:
Hi pentagonik, Thank you (or who else is taking care about all this audio issues) very much for doing this job. I also done an audio test on my other Windows 8 host / Windows 7 guest combination, so here are the audio files and logs. Maybe this could be useful for your debugging activities.
by , 6 years ago
Attachment: | VirtualBox-5.2.25-128361_Win78_log.7z added |
---|
5.2.25 logs, windows 7/8 system
comment:73 by , 5 years ago
I give a report about the current situation of this issue. Good news first: since the early betas of 6.0.5 and 5.2.27 everything works fine. In the 6.0.6 version I can’t hear anything wrong, in the 5.2.28 on my older Windows 8 host with Windows 7 guest system one can hear a small (0,1 second?) glitch after 4 minutes, which repeats itself approximately every 2 minutes after that. Probably I noticed that because I’m testing and aware. All currently supported versions working fine on my Windows 10 guest and host system. So in this corner of the audio system everything is fine.
Now to my other issue — switching the sound device of the host. It looks a little bit better now, but still not as good as in 5.1, where changing the output device of the host automatically causes VM to send the output into that direction. Current state: of course the complete shutdown and start helps as in 5.2.26 and 6.0.4. Restart of the VM or saving and restoring the state of the VM also helps now. It is sometimes enough just to click the cross in the toolbar of the VM (I’m in the full-screen mode) so the dialog asking what to do appears, then just cancel this dialog and the sound appears in the other device. Sometimes one have to repeat this couple of times to get the result.
Sometimes it helps to turn the audio output in the VM menu off and on, also couple of times. This worked in some early betas (6.0.5 and 5.2.27; 128442/3 and 128620/1), then disappeared, but now looks like working again. What still not working beside the automatic switch: turning off and on the playback in the VM. It also always helped the version 5.2 in the past, but not worked since the changes in the sound system during last fall.
I would try to make some log and audio files with extra audio debugging. In the past the change of the devices with active audio debugging caused the VM to stuck; after that the VM has had to be killed in the task manager of the host. It probably caused by the audio files writing routine. It would be probably better if we could suppress the writing of this files by some extra configuration. Then we would have at least more detailed logs itself. Of course if would be even better if this hanging in case of audio debugging and host device switching would completely disappear.
comment:74 by , 5 years ago
I created the logs and the audio files in audio debug mode. As I thought the change of the host audio output device cause the total hang of the VM, so I have had to kill it from the host’s task manager.
VBox_Audio_5229 VBox_Audio_607
Another question: this ticket was opened over the year ago for the different version of Virtual Box and for the general audio issues there. Those issues are nearly solved; what I’m writing here about is an audio problem, but another one. Shouldn’t I open another ticket for the current version of Virtual Box only for this sub-issue?
comment:75 by , 5 years ago
Tested VirtualBox 5.2.32 on Windows 7 host and Windows XP/7 guests.
Audio quality is good. There is no distortion/skipping/stutter. However, delay is built up while playing sound. Audio is in sync at start but drifts gradually. There is a 2 second delay after 2 hours of playing. It is noticeable when start, pause or resume playing. Disabling and enabling audio output twice can get audio in sync again.
Disabling audio output sometimes causes crackles. It also slows down playback timer on Windows 7 guest and stops playback timer on Windows XP guest.
comment:76 by , 5 years ago
I‘ve tested this issue with the latest builds from the 5.2, 6.0 and 6.1 branch. Generally the problem with jerking or lost audio output after the change of audio output device on the host during the playback persists. In the 5.2 version you could do this only once, if one tries to go back to the first audio output device on the host, the issue begins. In the 6.0 and 6.1 builds one can do it twice (forth and back), but then you have this troubles. Even if one manages to get sound on the right device, it still jerks there, even after the playback in the guest was stopped.
It only happens if one forgets to stop the playback before the transition and do the transition during the playback. The transition succeeds if it doesn't happen during the playback. What I do to get the sound into the right device is as usual following: restart the playback, turn the audio output off and on in the VM menu, click the cross in the same menu and cancel the appearing dialog about saving VM’s state and so on, saving and restoring the state. The only thing that helps right now is shutdown and start of the guest, not even the restart. The jerking sound continues even if the playback or audio output is off in the guest.
Finally, I managed to get the debug audio files without guest hanging during the transition in the debug mode. The trick is to do the first transition (speakers to headphone or Bluetooth) not during the playback, but stop the playback before. This way one has no immediate hangup of the guest and one can do as many transitions during the playback afterward as it wishes. So right now I got all the audio files and the logs from the audio debug session from all mentioned versions. The only problem: the audio files don’t reproduce the issue. They are all clear except of the places where I turned the audio off in the VM’s menu. There just the skip of the sound happens. So it clearly not the problem in the guest, it is only the problem between the guest or VBox and the host audio system.
5.2.33_r133385.7z 5.2.33_r133577.7z 6.0.13_r133347.7z 6.0.13_r133568.7z 6.1.0_BETA1_r133315.7z 6.1.0_BETA1_r133421.7z 6.1.0_BETA1_r133583.7z
by , 5 years ago
Attachment: | VBox_logs_29_sep_2019.7z added |
---|
Logs from current 5.2, 6.0 and 6.1 versions
comment:77 by , 5 years ago
I uploaded more fresh debug logs and sound files. Somebody added some additional sound logging into the source code; I saw nothing of it in the newly created logs but probably somebody could use this information anyway.
temp_5.2.35-134035.7z temp_6.0.15-134020.7z temp_6.1.0_BETA1-134041.7z
comment:78 by , 5 years ago
I tested with the currently available versions (5.2.35-135458, 6.0.15-135471, 6.1.1-135516). Sadly, but as expected, the issue with lost sound after switching the output devices on the host is still here. Nothing changes in the last months, even the newly available version 6.1.1-135516 shows the same behavior, which I already described enough in the last 16 month.
I put some logs and a link to the recorded files. The files and the complete log are only from the version 6.1.1-135516. In all other versions once again guest hangs completely if one switches the host audio output device in debug mode. It doesn't matter in which direction the switch happens, the only thing which triggers this bug is that the switch is happened during the running audio playback in the guest. Therefore, the logs from 5.2.35-135458 and 6.0.15-135471 are probably the partial ones.
comment:79 by , 5 years ago
I’ve tested my issue (sound lost after switching host audio output device) on the new betas (6.1.5 136449, 6.0.19 136450, 5.2.39 136276). In the 6.1 beta things looks a bit better. Sometimes it’s enough to stop and start the playback in the guest again. If this didn’t help, I still need to turn the sound off and on again in the VM menu or try to close the VM and after that cancel the save state dialog and go back to the VM. The 5.2 beta also looks a bit better, even if one needs more movements to solve the situation here. The 6.0 beta is the only one where you can see no changes since last builds.
In the 6.1 beta I even could switch the devices forth and back in debug mode. So for this version the logs and files are complete. Other versions still hangs after the first switch from one audio device to another.
So now things looking better and slightly moving in the direction of the version 5.2.22, which is the last one where one could switch the devices without a headache.
temp_6.1.5-136449.7z temp_6.0.19-136450.7z temp_5.2.39-136276.7z
by , 5 years ago
Attachment: | VBox_logs_16032020.7z added |
---|
Logs of 3 betas from the test on the 16. March 2020
comment:80 by , 4 years ago
I progressively got more and more audio problems with every update. the more I threw cores at the VM, the worse it got(16). Youtube is the worse ( i only use Firefox ). In any case got better results when I stayed at two cores, but still crackling a little. Hope it helps to find the issue. host windows 8.1 z820 2xE5-2687w HT enabled 32gb quad channels, VM Win7 64bits 8gb ram, emulation Intel HD Audio.
comment:81 by , 4 years ago
Take a look at the #18594 and the version 6.1.17 build 141128. This one looks like to be solved.
VBox.log zipped