TimeWarp Creates Rainbarrel and Granularity Noise

GoldWave general discussions and community help
Post Reply
djm
Posts: 2
Joined: Thu Apr 08, 2010 1:08 pm

TimeWarp Creates Rainbarrel and Granularity Noise

Post by djm »

I purchased GoldWave because I could not load my old Cool Edit Pro onto Win7. I want to be able to slow down a track, but keep the pitch the same in order to learn music.

When I use the TimeWarp effect, either Similarity or FFT, there is so much granularity noise and rain barrel effect as to make the music uninterpretable. I cannot find any information as to the significance of any of the settings in TimeWarp. How can I slow the music down, maintain pitch, but get the playback to be clear enough to be able to actually hear the music?

Thx,

djm
DougDbug
Posts: 2071
Joined: Wed Feb 16, 2005 3:33 pm
Location: Silicon Valley

Re: TimeWarp Creates Rainbarrel and Granularity Noise

Post by DougDbug »

Your results should be best with the FFT option. I assume you clicked on help in the Time Warp window? There are some hints for getting the best quality...

You can get artifacts with FFT processing, but I think you should be able to get "usable" results since you're not looking for "production quality".

As you probably know, Cool Edit was purchased by Adobe several years ago and is now Adobe Audition ($350 USD) and it does run on Win7. (I'm sooooo happy I chose GoldWave over Cool Edit, and I've been getting free upgrades for many-many years! ;) )

Audacity (FREE!!!) also has tools for changing tempo without altering pitch, so it might be worth a try too.
djm
Posts: 2
Joined: Thu Apr 08, 2010 1:08 pm

Re: TimeWarp Creates Rainbarrel and Granularity Noise

Post by djm »

Yes, I have played around with the settings in FFT, but was looking for some sort of explanation of the settings so I could eliminate the rain barrel effect.

I am aware of the travesty of Adobe ruining an excellent tool like Cool Edit, and I certainly won't blow $350 on something I already own.

Does Audacity slow down playback without introducing all these side effects?

Thx,

djm
mathyou9
Posts: 70
Joined: Mon Nov 10, 2008 7:36 pm

Re: TimeWarp Creates Rainbarrel and Granularity Noise

Post by mathyou9 »

I'm assuming that since you are trying to learn play the piece, you want to decrease the tempo as much as possible (which, for a single edit change, would be a 50% decrease in speed.)

I personally prefer to use Similarity when I need to change speed. However, I find that using the values that GoldWave suggests (i.e., "For music, larger Window size and Search range values give better results, such as 100 and 25") does, indeed, produce the "rain barrel" effect you you're experiencing. Instead, I find that using Similarity with the values 20 and 20 will produce decent-enough results to sufficiently learn-by-ear, keeping the rain barrel effect to a minimum.

Classical Gas (unchanged)

Classical Gas (50% decrease)

Although there is a very very slight rain barrel effect, most "common folk" would think the above altered piece was pretty transparent. Each picked note in my example is very discernible to make out and play (there aren't any "duplicate rain barrel" notes that arise from using the larger Window size.)

Good luck!
DougDbug
Posts: 2071
Joined: Wed Feb 16, 2005 3:33 pm
Location: Silicon Valley

Re: TimeWarp Creates Rainbarrel and Granularity Noise

Post by DougDbug »

I think you just have to experiment with the settings. More overlap and bigger FFT sizes should be better, but there are trade-offs and I don't know what's causing the "rain barrel" effect. The oscillator synthesis option is also supposed to give better results.

I only use Audacity occasionally and I've never used that feature. There are lots of optional plug-ins (both free and commercial) that work in Audacity, so if the built-in effect doesn't do the trick you can try searching for something better.

(I believe all of these programs will use an FFT based effect.)

Something you could try (with GoldWave)... Use the rate option to reduce the tempo and pitch together. Then use the Pitch effect to restore the pitch. I'm sure this uses the same FFT "engine", but the results might be different.
DewDude420
Posts: 1164
Joined: Fri Mar 11, 2005 11:15 pm
Location: Washington DC Metro Area
Contact:

Re: TimeWarp Creates Rainbarrel and Granularity Noise

Post by DewDude420 »

FFT is VERY confusing. It really is...I mean you're working on the audio in frequency space....which is a hell of a trick.

Furthermore, FFT resolutions and such can be confusing, and you wonder why you get these effects in the first place. It should be noted that FFT processing isn't perfect, and there are tradeoffs.

FFT Window basically says how many samples the algorhitm is going to use to make each processing section. Goldwave is one of the unique programs that gives the values as a power of 2 rather than the actual sample size, but I'm mostly just telling you that for informational purposes.

So basically what happens is if you have an FFT window size of say 1024 samples, then what's going to happen is 1024 samples of audio are going to be converted in to an FFT window for processing...then the processing is going to be applied, usually overlapping the FFT windows for more accurate processing and then spitting it out. This comes with a price though. The more samples you have per FFT window, the better frequency response you'll get at the cost of "time-resolution"..which means the filter won't respond quite as quickly because it's processing a larger number of samples at once. The fewer samples you use, you get much better time resolution but your frequency resolution goes down the crapper. You're either processing blocks of time or blocks of frequencies, neither one gives you a very accurate window. Things like overlap help...and the process is too difficult for me to explain right now (plus I haven't got the time...Fios guy is supposed to show up).

So, basically the rainbarrel and granularity noise are just results of lack of resolution either over time or in frequency during processing...and also the results of modifying in the frequency domain. USUALLY I get questions regarding those artifacts when people play with noise reduction, because the defect is about the same..and more pronounced since you are actually removing frequencies.
Post Reply