PDA

View Full Version : Lossy after pitch correction


Luap
2010-03-01, 03:35 PM
Hello everyone here.
I got one track (Holidays by the Beach Boys) from a bootleg. The pitch was off so I applied pitch correction in Sound Forge. When checking these files in Audiochecker, the original track checks out as CDDA sourced, but the pitch corrected track as MPEG sourced!
How can this be? Does the pitch correction really alter the quality of the audio, or is this a 'bug' in Audiochecker?
Thanks,
Paul

paddington
2010-03-01, 05:01 PM
It absoluetly alters it. That may be triggering the check to show positive - or you may have unknowingly re-encoded it when you saved it.

For sure, though, those auto-checkers are wrong all the time, so take it with a grain of salt.

anazgnos
2010-03-01, 05:22 PM
I think you've posted a couple times about false positives in audiochecker, and it can only lead me to conclude that audiochecker is entirely unreliable about guessing whether a track is actually lossy. Better to check this stuff out yourself with a spectral analyzer.

Of course pitch correction will "alter" the audio in a file but there's no reason why it should cause it to "lose" any high frequency information.

paddington
2010-03-01, 05:45 PM
I think you've posted a couple times about false positives in audiochecker, and it can only lead me to conclude that audiochecker is entirely unreliable about guessing whether a track is actually lossy. Better to check this stuff out yourself with a spectral analyzer.

Of course pitch correction will "alter" the audio in a file but there's no reason why it should cause it to "lose" any high frequency information.

The lossy data compression has nothing to do with high frequencies. Some algorithms for making MP3s, etc, also apply a low-pass roll-off, but not all. iTunes, for example, has a fantastic MP3 encoding process, when set correctly, that doesn't roll off (well, maybe a hair below 20k).. you can still tell it is lossy by the abrupt chunks missing from the spectrum. That is still the best way to tell and I've never seen an automatic method that is fool-proof for that.

Luap
2010-03-01, 05:55 PM
OK, thanks everyone for the answers. Audiochecker doesn't seem to be too reliable!
Can you guys recommend some good (free) software with a Spectral Analyzer? (I already searched on the internet but I didn't find a suitable one...)

paddington
2010-03-01, 05:57 PM
don't think there is... some of the stuff people post here using tegh free ones are just awful.. eye candy, not any good for analysis.

the one included with Cool Edit Pro 2.x (Adobe Audition) is great.. best I've seen in consumer software.

AAR.oner
2010-03-07, 08:23 AM
Audacity is about the best as far as freeware goes

PowerWindows
2010-03-13, 07:33 PM
Audacity is about the best as far as freeware goes

Yes, Audacity will do the trick.

http://audacity.sourceforge.net/

lordsmurf
2010-03-21, 09:02 PM
I'd suggest this is one area where Audacity doesn't really excel. Try SoundForge for this exact editing/restoring issue.

freezer
2010-05-20, 12:02 PM
Welcome to MyECStar.com - We are a well established Wholesale Supplier of Quality Fashion Shoes, Footwear and Accessories. Our Long Lasting Relationships with Several Large Quality Shoe Manufacturers, allows us to Consistantly provide our Customers with a Wide Variety of Top Quality Fashion Shoes, Footwear and Accessories. These Stylish and Trendy Fashion Footwear products are all offered at the lowest Competitive Wholesale Prices...Be sure and visit our website today....we have new styles of Fashion Productsarriving Weekly!

wowser.....what trendy footwear, but shouldn't this sort of spam be in the Lounge instead?

Certainly this may be just the type of fashions some of the TTD staff prefer, but how does this fluffy-stuff help sort out the lossy mp3 that eventually become silver pressed bootlegs that make up much of the ABT offerings?

just askin' ....... :lol4:

freezer
2010-05-20, 12:16 PM
hey, the fancy footwear done disappeared..........


:lmao:

rnranimal
2016-02-25, 04:52 PM
This is an old thread, but I can explain what happened. I will bet anything the speed correction made was in order to slow down the audio. When you slow the audio down, you reduce the pitch and thus the frequencies get lowered. In digital audio, you have a defined cutoff point. CD would be 22.05khz. So by whatever percentage you reduce the pitch/speed, you also reduce the frequencies. That leaves a gap up top, which would look like MPEG to testing software.

If you speed up the audio, you would be raising the frequencies and unless you raise the sampling rate, you would be loosing some upper frequencies as they will be above the upper cutoff. That's why instead of using speed/pitch processes in audio software, I adjust sampling rate tags without altering the audio to make speed corrections. Just as an example- If you take a 44.1 recording and it needs a 4% increase, you'd change the samplerate tags without altering the audio (Audacity can do this) to 45864khz. All this does is change the rate the audio is being played back and increases the samplerate (or decreases if slowing down). At this point, you have a 45864khz file. You can either resample back to 44.1 and loose some of your gained upper frequencies or upsample to 48khz and retained them (but now have a gap which might look like MP3).

rnranimal
2016-02-25, 05:14 PM
Just to add something… when I say changing the samplerate tag doesn't alter the audio, I meant it is a completely reversible process. If you change the 44100 tags to 45864 and then back to 44100, you end up back to the original 44100 file. It's just changing the rate in which the file is telling the software to play it at. When I make adjustments, I always keep a copy like this- say the 45864 file. Because you can always go back and make further adjustments without changing the actual audio. There is no need to increase bit-depth with this process. All you would be doing is adding blank bits at the bottom and maybe even dither noise if your software is setup to do that (as I believe Audacity is by default). You want to keep the audio at the original bit-depth and certainly not add any dither. If you do it right, the tags can always been changed back to the 100% original file. If audio checksum isn't the same as the original file after doing this, you did something wrong.

rnranimal
2016-02-25, 05:33 PM
This is an old thread, but I can explain what happened. I will bet anything the speed correction made was in order to slow down the audio. When you slow the audio down, you reduce the pitch and thus the frequencies get lowered. In digital audio, you have a defined cutoff point. CD would be 22.05khz. So by whatever percentage you reduce the pitch/speed, you also reduce the frequencies. That leaves a gap up top, which would look like MPEG to testing software.

If you speed up the audio, you would be raising the frequencies and unless you raise the sampling rate, you would be loosing some upper frequencies as they will be above the upper cutoff. That's why instead of using speed/pitch processes in audio software, I adjust sampling rate tags without altering the audio to make speed corrections. Just as an example- If you take a 44.1 recording and it needs a 4% increase, you'd change the samplerate tags without altering the audio (Audacity can do this) to 45864khz. All this does is change the rate the audio is being played back and increases the samplerate (or decreases if slowing down). At this point, you have a 45864khz file. You can either resample back to 44.1 and loose some of your gained upper frequencies or upsample to 48khz and retained them (but now have a gap which might look like MP3).

And just to be clear about that last part. When I am talking about resampling to 44.1 or 48, I am talking about a standard samplerate conversion which alters the audio. That would leave the speed correction in place, but alter the samplerate to your desired final spec. This is not reversible and is why I mention above that I would also keep a copy at the speed corrected samplerate (45864khz in my example).

I will stop talking to myself now. Well, on here at least. : )

trustthex
2016-05-29, 09:04 PM
No worries. You bumped a thread trolled by Art... and your answers totally make sense. thanks!