SonsOfMethuselah
All
I'm using VDJ 5.0 Rev 6 on an Intel Dual Core 7208 1.8Ghz Laptop with a 120Gb SATA HDD. I'm running XP SP2 and the machine has a Realtek HD Audio output onboard. I'm using a BCD3000 unit with the ASIO driver as my main sound device for VDJ.
I've managed to resolve all my previous issues looking through the forums and following all the tricks and tips for optimising windows itself for performance (e.g. background task priority, system optimize for performance, registry tweeks etc) and also I've played around with all the VDJ settings for latency, performance, algorithms etc etc but NONE of them resolve this issue.
What I'm hearing is a very slight time delay in the playback of mp3 files, and I don't mean the latency of response to any key commands or between the BCD unit and reaction from VDJ, I mean when I'm listening to a file playing, I can liken it to almost an LFO effect, except with the actual time of the playback rather than any frequency of the sound. Everything is just not quite on beat!
Now I do stress this is "slight" probabably < 5-6ms, but enough if you are listening and trying to beat match is can bug the sh*t out of you. The other thing I can add, is this does appear to be more noticable the higher the bitrate of the file is, i.e. a file encoded at 128k you cannot hear the delay at all, but at 320k it's as described above.
I also have an external USB soundcard and I have tried using all 3 outputs as the primary output for VDJ, the same issue happens on all 3. However I can play all of these files using winamp, media player, quicktime etc and there is no audible beat delay at all on any of the output cards. It's almost like the realtime decoding of the mp3 file in VDJ is struggling, yet the processor never goes anywhere near the top utilisation.
Any suggestions that could help me eliminate this would be most welcome.
Thanks
Denzel
I'm using VDJ 5.0 Rev 6 on an Intel Dual Core 7208 1.8Ghz Laptop with a 120Gb SATA HDD. I'm running XP SP2 and the machine has a Realtek HD Audio output onboard. I'm using a BCD3000 unit with the ASIO driver as my main sound device for VDJ.
I've managed to resolve all my previous issues looking through the forums and following all the tricks and tips for optimising windows itself for performance (e.g. background task priority, system optimize for performance, registry tweeks etc) and also I've played around with all the VDJ settings for latency, performance, algorithms etc etc but NONE of them resolve this issue.
What I'm hearing is a very slight time delay in the playback of mp3 files, and I don't mean the latency of response to any key commands or between the BCD unit and reaction from VDJ, I mean when I'm listening to a file playing, I can liken it to almost an LFO effect, except with the actual time of the playback rather than any frequency of the sound. Everything is just not quite on beat!
Now I do stress this is "slight" probabably < 5-6ms, but enough if you are listening and trying to beat match is can bug the sh*t out of you. The other thing I can add, is this does appear to be more noticable the higher the bitrate of the file is, i.e. a file encoded at 128k you cannot hear the delay at all, but at 320k it's as described above.
I also have an external USB soundcard and I have tried using all 3 outputs as the primary output for VDJ, the same issue happens on all 3. However I can play all of these files using winamp, media player, quicktime etc and there is no audible beat delay at all on any of the output cards. It's almost like the realtime decoding of the mp3 file in VDJ is struggling, yet the processor never goes anywhere near the top utilisation.
Any suggestions that could help me eliminate this would be most welcome.
Thanks
Denzel
Posté Sat 29 Dec 07 @ 6:56 pm
Paz75
what are you using to output your sound? so you're saying that the latency is post fader (after the main output)?
I had a similar problem but that wasnt my computer, it was the reciever i plugged my mixer mains into as it was a surround mixer. i just had to figure out how to turn surround sound off.
essentially there are all sorts of BS surround sound modes which arent surround at all. all it does is put a delay and a frequency filter on different channel outputs. even the front speakers were delayed a few ms enough to make mixing uncomfortable.
I had a similar problem but that wasnt my computer, it was the reciever i plugged my mixer mains into as it was a surround mixer. i just had to figure out how to turn surround sound off.
essentially there are all sorts of BS surround sound modes which arent surround at all. all it does is put a delay and a frequency filter on different channel outputs. even the front speakers were delayed a few ms enough to make mixing uncomfortable.
Posté Sun 30 Dec 07 @ 4:50 am
SonsOfMethuselah
Hi Paz
I'm just using this setup at home at moment so this is a straightforward aux channel input into my old awai stereo system. Dont think they even had surround sound when this thing was built!
Nevertheless I've just tried it into my Line6 guitar amp with all effects off, and my Sony DB930 5.1 amp, again with all effects off and it still sounds the same in both.
I've just experimented with it a bit more, and actually this is happening if I key lock the track and change the bpm. The track I tried it on is 128bpm originally, if I nudge it up to 130bpm, you can hear the lag kicking in. I've also since come across an old thread that suggested the mp3 decoding engine could be a bit suspect. So I've even tried ripping the same song from a CD straight to wav file, 16bit 44khz. No difference. As soon as the song moves off it's original bpm at all, you can hear this. However, with the keylock off, it's happy to speed up and slow down, even by much larger margins of bpm and it seems to be ok.
So I guess it must be something to do with the realtime transformations taking place when you change the tempo of the song, but not the key, or vice versa (I slowed it down and locked the key, then reset the bpm, sounds even worse!).
This is also a bit of a pain as it's mean I can't beatmatch and keymatch at the same time!
I'm now wondering if I am actually hearing a delay, or if in the keyshift process some kind of "flanging" of the output is occuring thats just giving the audible effect of the main beat being delayed. I remember from messing with soundforge and the like in earlier days if you pitch shifted a song you would get some flanging and frequency oscillation depdending on the algorithm used to shift it.
Maybe it's one for the technical gurus to look at.
Thanks
Denzel
I'm just using this setup at home at moment so this is a straightforward aux channel input into my old awai stereo system. Dont think they even had surround sound when this thing was built!
Nevertheless I've just tried it into my Line6 guitar amp with all effects off, and my Sony DB930 5.1 amp, again with all effects off and it still sounds the same in both.
I've just experimented with it a bit more, and actually this is happening if I key lock the track and change the bpm. The track I tried it on is 128bpm originally, if I nudge it up to 130bpm, you can hear the lag kicking in. I've also since come across an old thread that suggested the mp3 decoding engine could be a bit suspect. So I've even tried ripping the same song from a CD straight to wav file, 16bit 44khz. No difference. As soon as the song moves off it's original bpm at all, you can hear this. However, with the keylock off, it's happy to speed up and slow down, even by much larger margins of bpm and it seems to be ok.
So I guess it must be something to do with the realtime transformations taking place when you change the tempo of the song, but not the key, or vice versa (I slowed it down and locked the key, then reset the bpm, sounds even worse!).
This is also a bit of a pain as it's mean I can't beatmatch and keymatch at the same time!
I'm now wondering if I am actually hearing a delay, or if in the keyshift process some kind of "flanging" of the output is occuring thats just giving the audible effect of the main beat being delayed. I remember from messing with soundforge and the like in earlier days if you pitch shifted a song you would get some flanging and frequency oscillation depdending on the algorithm used to shift it.
Maybe it's one for the technical gurus to look at.
Thanks
Denzel
Posté Sun 30 Dec 07 @ 6:52 am
Paz75
do you have a monitor on your input? i get flanging when i input from my mixer which is in turn monitored back through the soundcard. it doesnt feedback like you would expect in the analog world, but it is a loop of sorts..
also have you tried just mixing in the headphones to see if the same occurs?
also have you tried just mixing in the headphones to see if the same occurs?
Posté Sun 30 Dec 07 @ 9:48 am
erxon
So basically what you are hearing is beat-skips (not time delay), like the kick is not punchy sometimes and/or it hits slightly off? That's the sound you will always get with current algorithm. Even if you use high quality algorithm settings, it will still be noticable now and then.
I made 3 samples so you can compare if this is what you are experiencing:
SAMPLE 1 (original recording)
SAMPLE 2 (VDj live-processed master tempo @ 3.9% stretching)
SAMPLE 3 (Cubase SX offline-processed advanced algorithm @ 3.9% stretching)
In sample 2, the sound is clean but not precise on time, while in sample 3 you get a correct timing but additional "wah"-like sound, if you listen carefully.
I made 3 samples so you can compare if this is what you are experiencing:
SAMPLE 1 (original recording)
SAMPLE 2 (VDj live-processed master tempo @ 3.9% stretching)
SAMPLE 3 (Cubase SX offline-processed advanced algorithm @ 3.9% stretching)
In sample 2, the sound is clean but not precise on time, while in sample 3 you get a correct timing but additional "wah"-like sound, if you listen carefully.
Posté Sun 30 Dec 07 @ 12:33 pm
SonsOfMethuselah
erxon
Yeah, sample two is exactly what I'm getting. I guess from your response you are saying you would expect the algorithm to do this? Is this some kind of technical limitation on the way the engine processes timestretching.
I'd be interested to know if this is particularly a windows version based problem or do mac's suffer from it too?
Interestingly I've now gone and whacked the performance presets up to best quality and low and behold it is much less noticable. I initially had problems with with VDJ at this setting though, but I've not tried it since doing all the performance tweeks to windows so I'll give it a go for a few days and see how I get on.
Many thanks for your help and time folks.
Denzel
Yeah, sample two is exactly what I'm getting. I guess from your response you are saying you would expect the algorithm to do this? Is this some kind of technical limitation on the way the engine processes timestretching.
I'd be interested to know if this is particularly a windows version based problem or do mac's suffer from it too?
Interestingly I've now gone and whacked the performance presets up to best quality and low and behold it is much less noticable. I initially had problems with with VDJ at this setting though, but I've not tried it since doing all the performance tweeks to windows so I'll give it a go for a few days and see how I get on.
Many thanks for your help and time folks.
Denzel
Posté Mon 31 Dec 07 @ 8:05 am
erxon
Yes, this effect is produced with the current timestretch algorithm that VDj uses. So even though I don't have a mac to experimentaly determine, I can safely say it is not platform dependant, as both versions should use the same algorithm. I suggest you do some testing with different Complexity and Spatialization settings for Advanced Master Tempo, and refer to these links for more info on settings and algorithm itself:
Master Tempo Manual Config
Spatialization and Master Tempo
www.dspdimension.com (DIRAC Audio Time Stretching and Pitch Shifting that VDj uses)
P.s.: Using higher settings can freeze low-end PCs, and it cannot handle very low latencies.
e
Master Tempo Manual Config
Spatialization and Master Tempo
www.dspdimension.com (DIRAC Audio Time Stretching and Pitch Shifting that VDj uses)
P.s.: Using higher settings can freeze low-end PCs, and it cannot handle very low latencies.
e
Posté Tue 01 Jan 08 @ 12:05 pm
Paz75
whats the compression rate of the mp3s you used erxon? does it do the same with 320? im just curious cuz ive never noticed this in vdj before. but you can definitley hear that the beat is not totally accurate
Posté Wed 02 Jan 08 @ 3:11 am
erxon
You're right Paz, I haven't checked the bitrate. That song was encoded to 192kbit/s, so I made another example
using a wave cd quality with one of my demos:
Original wav recording [128 bpm converted to mp3 320kbit/s]
Timestretched wav recording [3.9% (133bpm) converted to mp3 320kbits/s after being recorded in wav quality]
This was made using Advanced Master Tempo algorithm with both spatialization and complexity params set to 1.
Do you get better results? Can you provide your example?
using a wave cd quality with one of my demos:
Original wav recording [128 bpm converted to mp3 320kbit/s]
Timestretched wav recording [3.9% (133bpm) converted to mp3 320kbits/s after being recorded in wav quality]
This was made using Advanced Master Tempo algorithm with both spatialization and complexity params set to 1.
Do you get better results? Can you provide your example?
Posté Wed 02 Jan 08 @ 5:58 am
Paz75
can you try it with a true 320? encoding up probably wont help since like a pic that's resized, you're doubling the existing pixels.
try ripping any cd track to 320 directly and then try... the resolution should work now.
try ripping any cd track to 320 directly and then try... the resolution should work now.
Posté Wed 02 Jan 08 @ 11:42 am
erxon
No, you missunderstood. This demo is actually produced with a resolution of 48khz & 24bits, diethered to 44.1/16, so it wasn't upsampled ;) I only downsampled to mp3 for listening on internet. So I should actually get better sound results than processing mp3s, however it isn't so.
Posté Wed 02 Jan 08 @ 12:18 pm
Paz75
whoops, sorry dude. i misread 128bpm sampled to 320kpbs as 128kbps sampled to 320.. mah bad.
sorry to be a bother but could you do one that isnt so jazzy? the syncopation is a bit hard to listen for discrepencies. i do have to say it does sound _slightly_ off, but much less off than the previous samples you had. do you agree?
i'll have a tinker on my side tonight with it. im also curious, cuz it seems that the higher rate you go, the problem lessens. maybe a regular wav exhibits no problem at all?
if that's true, i wouldnt say it's 'fine' but that it does raise an interesting mathmatical problem...
sorry to be a bother but could you do one that isnt so jazzy? the syncopation is a bit hard to listen for discrepencies. i do have to say it does sound _slightly_ off, but much less off than the previous samples you had. do you agree?
i'll have a tinker on my side tonight with it. im also curious, cuz it seems that the higher rate you go, the problem lessens. maybe a regular wav exhibits no problem at all?
if that's true, i wouldnt say it's 'fine' but that it does raise an interesting mathmatical problem...
Posté Wed 02 Jan 08 @ 9:57 pm
djcel
In order to provide a better quality, the advanced master tempo has to work on a bigger sound buffer. Consequently it adds a latency in the signal.
If you want to avoid this latency, use the fast master tempo which works with low sound buffers.
If you want to avoid this latency, use the fast master tempo which works with low sound buffers.
Posté Fri 04 Jan 08 @ 5:50 am
erxon
Paz75 wrote :
whoops, sorry dude. i misread 128bpm sampled to 320kpbs as 128kbps sampled to 320.. mah bad.
sorry to be a bother but could you do one that isnt so jazzy? the syncopation is a bit hard to listen for discrepencies. i do have to say it does sound _slightly_ off, but much less off than the previous samples you had. do you agree?
i'll have a tinker on my side tonight with it. im also curious, cuz it seems that the higher rate you go, the problem lessens. maybe a regular wav exhibits no problem at all?
if that's true, i wouldnt say it's 'fine' but that it does raise an interesting mathmatical problem...
sorry to be a bother but could you do one that isnt so jazzy? the syncopation is a bit hard to listen for discrepencies. i do have to say it does sound _slightly_ off, but much less off than the previous samples you had. do you agree?
i'll have a tinker on my side tonight with it. im also curious, cuz it seems that the higher rate you go, the problem lessens. maybe a regular wav exhibits no problem at all?
if that's true, i wouldnt say it's 'fine' but that it does raise an interesting mathmatical problem...
I'll add one sample that is only kick and simple rythm later on, however I cannot agree that latest sample was less off then the first one. They both sound unacceptably off for professional use. In theory you are correct, higher sampling & bit rates will naturally give you better results after processing (losing data), but it seems I cannot get it to sound at least so rythmically correct as I can do with an mp3 in cubase sx (whatever latency option I choose).
Ps.: I'm not sure what exactly you mean by "regular wav". The last sample I used was by all means a regular wav. :)
Posté Fri 04 Jan 08 @ 6:56 am
Paz75
Ok, here is my submission. I have two track placed, both have 3 tests in order
1) regular pitch (124bpm)
2) pitched up (134bpm)
3) pitch locked (124bpm)
The two files in order are Mp3 source at 320kbps and WAV
Both are compressed to 320 afterwards. For me, I only really detect aliasing in the pitch locked versions and only detect slightly less aliasing in the WAV version, and thats only before reencoding the recording to 320. After encoding, you dont hear any differences at all.
Both seem acceptable to me for playing live.
mp3_320_test
http://www.djpaz.net/images/mp3_test.mp3
wav_test
http://www.djpaz.net/images/wav_test.mp3
1) regular pitch (124bpm)
2) pitched up (134bpm)
3) pitch locked (124bpm)
The two files in order are Mp3 source at 320kbps and WAV
Both are compressed to 320 afterwards. For me, I only really detect aliasing in the pitch locked versions and only detect slightly less aliasing in the WAV version, and thats only before reencoding the recording to 320. After encoding, you dont hear any differences at all.
Both seem acceptable to me for playing live.
mp3_320_test
http://www.djpaz.net/images/mp3_test.mp3
wav_test
http://www.djpaz.net/images/wav_test.mp3
Posté Sat 05 Jan 08 @ 10:49 pm
erxon
Nice one paz, your samples actually sound better even at 4% higher stretching. It still sounds slightly off tempo (enough that it annoys me :P). What settings of spatialization and complexity are you using? I'll try recording another one with simple beat tomorrow, after I get rid of the weekend hangover.
Posté Sun 06 Jan 08 @ 9:34 am
Paz75
Here's my settings:
I'll have a play with the settings today probably and see if I can get any better results. I'm using vdj along with orion and ableton to do some old jazz remixes and desire the best quality i can get. The steps are
1) pitch shift to an approximate speed in vdj and bounce
2) loop and beat correct in ableton and export to samples
3) arrange in orion
Works quite well, got some results here : http://www.djpaz.net/index.php?option=com_xemusicfx&func=detail&id=15
Anyhow, you can expect that pitch shifting will not sound as good. As I understand it (which could be incorrect) that pitch shifting relies first on an accurate estimation of the tempo and ccg placement. the second step is that slices are taken in conjunction to the tempo and each section is applied with a pitch shift algorythmn and then fed back through the tempo engine. the result at best is an amount of warbling which you get from the individually applied segments. but it seems to work well enough for me
I'll have a play with the settings today probably and see if I can get any better results. I'm using vdj along with orion and ableton to do some old jazz remixes and desire the best quality i can get. The steps are
1) pitch shift to an approximate speed in vdj and bounce
2) loop and beat correct in ableton and export to samples
3) arrange in orion
Works quite well, got some results here : http://www.djpaz.net/index.php?option=com_xemusicfx&func=detail&id=15
Anyhow, you can expect that pitch shifting will not sound as good. As I understand it (which could be incorrect) that pitch shifting relies first on an accurate estimation of the tempo and ccg placement. the second step is that slices are taken in conjunction to the tempo and each section is applied with a pitch shift algorythmn and then fed back through the tempo engine. the result at best is an amount of warbling which you get from the individually applied segments. but it seems to work well enough for me
Posté Sun 06 Jan 08 @ 9:19 pm
erxon
I tried with your settings and got this:
Original
Stretched to 4%
Stretched to 8%
And 4% actually sounds just ok. I rarely go +3% anyway so it's good enough.
Nice style, I'm also quite into remixing/mimicking old jazz & swing stuff.
Yes it works in that way, and as long as you stick to small changes from original it will be acceptable.
The VDj CBG (beat grid) is only visual representation, so it does not play any role in timestretch process.
Original
Stretched to 4%
Stretched to 8%
And 4% actually sounds just ok. I rarely go +3% anyway so it's good enough.
Paz75 wrote :
Works quite well, got some results here : http://www.djpaz.net/index.php?option=com_xemusicfx&func=detail&id=15
Nice style, I'm also quite into remixing/mimicking old jazz & swing stuff.
Paz75 wrote :
Anyhow, you can expect that pitch shifting will not sound as good. As I understand it (which could be incorrect) that pitch shifting relies first on an accurate estimation of the tempo and ccg placement. the second step is that slices are taken in conjunction to the tempo and each section is applied with a pitch shift algorythmn and then fed back through the tempo engine. the result at best is an amount of warbling which you get from the individually applied segments. but it seems to work well enough for me
Yes it works in that way, and as long as you stick to small changes from original it will be acceptable.
The VDj CBG (beat grid) is only visual representation, so it does not play any role in timestretch process.
Posté Tue 08 Jan 08 @ 4:06 am
Groove D
i had the "wah wah" warblinh effect like a table with a loose belt. I turned off KEYLOCK ( the little padlocks in the bottom left and right of each deck) and like magic things play realtime and dont fight each other trying to syncronize
try that
try that
Posté Sun 20 Jan 08 @ 4:41 am