Slight inaccuracy in playback speed

[Sentenza]

I noticed what looks like a slight inaccuracy in the playback speed that seems proportional to the ratio between the track's base speed and the playing speed.

For ex, let's say I'm playing a track at 125 bpm ind deck A, I load another track with base at 126 bpm in deck B and hit the sync button, it will first look like it's synced but a delay will appear progressively (this is also visually seen with the beat markers). If I now load another track @126BPM in deck A and sync, they will play perfectly in sync.

The more there is difference between base speed and playing speed, the more I get delay.

System: mixxx 1.8.0 beta1, jack with firewire soundcard, ubuntu karmic 64bits , intel Q6600.

[Albert Santoni (gamegod)]

Hi Sentenza,

Does this occur if you use both Pitch-Independent Time Stretch and
Vinyl Emulation mode in the preferences?

Thanks,
Albert

On Tue, Apr 6, 2010 at 9:11 AM, Sentenza <email address hidden> wrote:
> Public bug reported:
>
> I noticed what looks like a slight inaccuracy in the playback speed that
> seems proportional to the ratio between the track's base speed and the
> playing speed.
>
> For ex, let's say I'm playing a track at 125 bpm ind deck A, I load
> another track with base at 126 bpm in deck B and hit the sync button, it
> will first look like it's synced but a delay will appear progressively
> (this is also visually seen with the beat markers). If I now load
> another track @126BPM  in deck A and sync, they will play perfectly in
> sync.
>
> The more there is difference between base speed and playing speed, the
> more I get delay.
>
> System: mixxx 1.8.0 beta1, jack with firewire soundcard, ubuntu karmic
> 64bits , intel Q6600.
>
> ** Affects: mixxx
>     Importance: Undecided
>         Status: New
>
> --
> Slight inaccuracy in playback speed
> https://bugs.launchpad.net/bugs/556639
> You received this bug notification because you are a member of Mixxx
> Development Team, which is subscribed to Mixxx.
>

[Sentenza (sentenzux)]

It actually occurred in vinyl emulation mode... Will check with PITS and post result.

[RAFFI TEA (raffitea)]

Hi all,

just bear in mind that BPM detection is an estimation, hence, SYNC' is only as reliable as your BPM algorithm is.

So are you sure that your base speed is *exactly* 126 BPM? Did you count the beats manually?

[Sentenza (sentenzux)]

I'm sure about the speed of the tracks tested (at least integer part of it ;) ), and the proportionality of the problem is really clear.

If i do the same test with a 130 bpm track, the problem gets just worst, so I don't think this is related to bpm detection.

[Sentenza (sentenzux)]

I definitely confirm this bug in vinyl mode. When track is played at pitch>100% , it plays slightly slower than expected, and slightly quicker when pitch < 100%. This is also visible in the waveform as beat marks will get unaligned.

This bug also do not seems to affect PITS mode, or with far less effect. However i noticed that the sound and waveform seem to get unsynchronized after some time. This particularly visible on house/techno tracks when beat start again after a beat-less period.

[Sentenza (sentenzux)]

Checked with beta2 yesterday, and this bug is still present in vinyl mode. PITS mode crashed almost immediately, so I can't confirm.
I suspect this may be related to the fact I'm playing at 48KHz, which is the native mode of my firewire soundcard...

[Sentenza (sentenzux)]

This seems definitely related to playback rate: playing at 48 KHz trigger the bug but not 44KHz. I'll try to check at 96, but didn't have time for now. My setup:

 - Ubuntu 10.04 x86_64 - Intel Q6600
 - Edirol FA-101 using ffado with jack running realtime
 - External Pioneer mixer

Also BPM are coherent between my DJM700 and mixxx when playing 44, but my mixer always measure ~1 BPM more when playing at 48 (I did not report before because I was not sure which one was right).

[Phillip Whelan (pwhelan)]

The latest version of trunk and 1.8 (from source) have a small fix for this. It should at stay synced for much longer than before.

[RJ Skerry-Ryan (rryan)]

Is this still a problem Phil?

[Albert Santoni (gamegod)]

This will definitely still be a problem if we're not keeping track of
the playback position in floats.... (my 2cents)

On Mon, Oct 18, 2010 at 2:51 PM, RJ Ryan <email address hidden> wrote:
> Is this still a problem Phil?
>
> --
> Slight inaccuracy in playback speed
> https://bugs.launchpad.net/bugs/556639
> You received this bug notification because you are a member of Mixxx
> Development Team, which is subscribed to Mixxx.
>

[KWhat (alex.barker)]

This problem is still occurring for me in the trunk. Deck B seemed to play ever so slightly faster. BPM markers are the same, but the bottom one just moves faster.

[Albert Santoni (gamegod)]

Based on Owen's test results, we believe this is fixed in the features_scratch2 branch and will ship in Mixxx 1.10.

Thanks,
Albert

[Sentenza (sentenzux)]

Hi guys,

I tested 1.9 beta yesterday, and got some more issues with playing speed. I quickly noticed that PITS and standard more were definitely not playing the track at same speed (at least mesured with BPM meter in my DJM700). PITS seems to be at the right speed but playback is strange (sound like the speed is regularly changed). Playback in standard mode is definitely not done at right speed.

I'll try to fetch features_scratch2 and test it.

Otherwise, 1.9 is a pretty impressive release... Good work guys!!!

[Sentenza (sentenzux)]

Is this already merged in trunk?

[Owen Williams (ywwg)]

No, not in trunk. features_xwax2 is the only place that has it. It would be trivial to move into trunk though, just minor changes to enginebuffer and major ones to enginebufferscalelinear.

[Swiftb0y (swiftb0y)]

Mixxx now uses GitHub for bug tracking. This bug has been migrated to:
https://github.com/mixxxdj/mixxx/issues/5378