Subject:Time Stretch

Posted by: Jayson H.

Date:5/16/2005 7:28:02 PM

Hi,
I'm trying to time stretch a piece of slow rock music in Sound Forge. I'm no audio expert and if I just do a basic time stretch in SF7 the result has some slight stutters in it... not really in the foreground(sounds like its in the bass) but still noticeable. The stretch is about 110% of the original time. The original is ripped from CD so the quality there should be clean.
I did a search and didn't find anything relevant... does anyone have any suggestions on how to make a time stretch as clean as possible? Or how to clean it up through filtering afterwards?
Thanks for any suggestions.

-Jayson

Subject:RE: Time Stretch

Reply by: Spheris

Date:5/17/2005 7:13:34 AM

Best suggestion is this, check and trim the mid/side signals so phase discrepancies are minimal. usually best average is left/mid channel should be -1 right/side channel should be around -12-16 before returning them to normal stereo.

second option is process each channel separately and restrip them to the original - pitch shifters do NOT do well with stereo signals at all

Subject:RE: Time Stretch

Reply by: Jayson H.

Date:5/17/2005 7:53:16 AM

Spheris,
I greatly appreciate your advice. Unfortunately, I'm not exactly clear what you are suggesting... it's a little too technical. I know what stereo is, I know what phase discrepancies in stereo signals are. What is the "left/mid" channel and the "right/side" channel, and how does that differ from the left channel and the right channel?

Forgive me if I'm really showing my ignorance here. If it will take too long to discuss, I understand. You can just point me to some documentation on it as well if you have any; I'm not afraid to read up a bit.

-Jayson

Subject:RE: Time Stretch > Vegas Stretch Attributes

Reply by: Douglas Clark

Date:5/18/2005 4:12:58 AM

Jayson

I don't have Sound Forge at hand, but Vegas offers 19 different methods for time stretch, if the default method leaves artifacts (doesn't sound good). They are referred to as Stretch Attributes under Event Properties. Maybe this is relevant to Sound Forge, too. In Vegas, I have stretched/compressed events using ctrl-drag on the edge of the event. I haven't tried it in SF.

(the following refers to Vegas5)

Select a section of your stretched event that sounds poor, and play it looping. Then right-click the event and select properties. The bottom half of the Audio Event tab shows the Time Stretch / pitch shift parameters. Click on the Stretch Attributes dropdown box, and then cycle through the choices using up and down arrow keys. Allow a second or two for the processing to switch to each new method. Stop when you find one that sounds OK. (If you stretch too much, none of them will sound good).

Then repeat/confirm the process on the whole event, because a new method may make one part sound better, but another part worse.

Under "Editing Event Properties", the Vegas5 help file says "Choose a setting from the Stretch Attributes drop-down list to specify how you want to divide and crossfade the file to prevent artifacts. Depending on your source material, you may need to experiment with different crossfade types."

I don't know what changing crossfade types will do, or if that is only relevant in Vegas. Anybody know more about how this works?

Douglas

Subject:RE: Time Stretch

Reply by: Jayson H.

Date:5/18/2005 11:18:32 AM

All,
Some further thoughts:

I'm pretty sure the problem is definitely in the bass. If I use the EQ to cut out bands 2 and 3 of the 10 band EQ, the stutter sound is not audible. I tried to separate the sound into high and low components using the EQ, filter out the low end with a little chorus to cover up the pops, and recombine the sounds, but that didn't seem to help a lot. I tried a bunch of different stretching modes and the "Solo Instrument" one works the best, but I'm not as satisfied as I'd like to be. I've tried all sorts of the included plug-ins (maybe there are ones on the net that could help?) to filter things out, but nothing I was happy with.

I also tried mixing the stereo signals into a mono signal and only applying stretch on the mono signal. The same artifacts were present so I don't think it has to do with phase issues in the stereo.

Any further thoughts?

-Jayson