Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: changing speed using Audacity (audible quality loss ?) (Read 12280 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

changing speed using Audacity (audible quality loss ?)

Hello,

If I was to change the speed of a 3-4 minute song, either slowing it down or speeding it up by about 5% using Audacity, and then changed it back to it's original speed, would there be any "audible" quality loss ?

changing speed using Audacity (audible quality loss ?)

Reply #1 –
Why not try it yourself?

changing speed using Audacity (audible quality loss ?)

Reply #2 –
The easiest route I can think of to achieve the same goal would be to not speed it up or slow it down by 5% and then not change it back to its original speed because it's already there. This would guarantee no quality loss.

changing speed using Audacity (audible quality loss ?)

Reply #3 –
Quote from: Slipstreem on
The easiest route I can think of to achieve the same goal would be to not speed it up or slow it down by 5% and then not change it back to its original speed because it's already there. This would guarantee no quality loss.


hahah YES. haha. Of course.
Just asking out of curiosity really....
I know it's a lossy process, but I don't really know to what extent you can change something before you can hear quality loss, so I was just wondering if there really is a definitive technical answer ?
I know when correcting the speed of old cassette tapes I'd never really need to change anything more than 5%, so thought I'd use that as an example, would changing backwards and forwards in a 5% range cause audible quality loss ?

changing speed using Audacity (audible quality loss ?)

Reply #4 –
I changed the speed on a vinyl rip I made which was about 4% out - using audacity, I was impressed at how well it came out.

changing speed using Audacity (audible quality loss ?)

Reply #5 –
My guess is no...  You won't hear a difference.  You are just re-sampling (interpolating & filtering).  I frequently re-sample (without changing speed) between 44.1kHz (CD) and 48kHz (video) and I never notice anything.  Of course I hate doing it because I "know" it's lossy...   

If you go "too far" in speeding-up, you run into "Nyqyist" and you'll loose high frequencies when you slow it back down.  (You could minimize that by up-sampling when you speed-up.)

It gets a lot more dangerous if you start changing speed while mantaining pitch or vice-versa.  That involves FFT and you are more likely to get audible artifacts.

Quote
I know when correcting the speed of old cassette tapes I'd never really need to change anything more than 5%, so thought I'd use that as an example, would changing backwards and forwards in a 5% range cause audible quality loss ?
I don't think I would notice a 5% change (without an A/B comparison), but if you have perfect pitch, or if you are a musician trying to play in-tune with the recording, a slight "quality loss" is preferable to pitch/tempo problems.


Quote
...so I was just wondering if there really is a definitive technical answer ?
It's hard to say "definitive technical" when we're talking about something that might be on the edge of human perception.   

changing speed using Audacity (audible quality loss ?)

Reply #6 –
Quote from: bomber on
I know when correcting the speed of old cassette tapes I'd never really need to change anything more than 5%, so thought I'd use that as an example, would changing backwards and forwards in a 5% range cause audible quality loss ?

Don't some cassette decks have speed adjustments? If your does then that would be the way to go.

With the amount of noise and distortion in tape, I would think that artifacts in digital resampling would be the least of your problems.

changing speed using Audacity (audible quality loss ?)

Reply #7 –
Quote from: pdq on
With the amount of noise and distortion in tape, I would think that artifacts in digital resampling would be the least of your problems.

With an expensive Type IV tape recorded in Dolby C on a professional quality deck with Dolby HX-Pro, I wouldn't want to place any bets on that.

changing speed using Audacity (audible quality loss ?)

Reply #8 –
Quote from: DVDdoug on
My guess is no...  You won't hear a difference.  You are just re-sampling (interpolating & filtering).  I frequently re-sample (without changing speed) between 44.1kHz (CD) and 48kHz (video) and I never notice anything.  Of course I hate doing it because I "know" it's lossy... 

Actually, if you slow something down, you are actually adding samples to the file and making it longer/larger.  If you speed something up, you are discarding samples to make the file shorter/smaller.  The sampling rate is not being changed.

Doing one, then the other, you will end up with the same number of samples as you had originally but they won't be exactly the same samples, hence the potential for an audible loss of quality.  That said, I don't know how to answer the question without testing and listening.

changing speed using Audacity (audible quality loss ?)

Reply #9 –
Quote
If you speed something up, you are discarding samples to make the file shorter/smaller. The sampling rate is not being changed.
  Yes....  And, when you downsample, you are discarding samples (which makes the file smaller), but in this case you change the file header so that the playback speed is correct.

changing speed using Audacity (audible quality loss ?)

Reply #10 –
Quote from: DVDdoug on
If you go "too far" in speeding-up, you run into "Nyqyist" and you'll loose high frequencies when you slow it back down.  (You could minimize that by up-sampling when you speed-up.)

When you speed it up, the highest frequencies are pushed beyond where you can hear and beyond what the digital audio format can represent (assuming you use the same sample rate for the sped-up version). When you slow it back down, you won't get them back.

The process here is called sample-rate-conversion which can, in theory and in practice, be transparent. It's a computationally intensive process and so in the interest of speed (or incompetence), some implementations take shortcuts and distortion is introduced.

changing speed using Audacity (audible quality loss ?)

Reply #11 –
Some info from the "AUDACITY" forum specifically on how the programme performs this process, it seems to me that it's extremely unlikely I will hear a difference if I change the speed back and forth by just 5% a couple of times :
----------
Yes, I do a lot of testing of Audacity. Yes Change Tempo and Change Pitch cause noticeable sound quality loss, particularly with large changes. Yes Change Speed causes a measurable distortion, but it is primarily "phase shift" which is virtually inaudible. There is also some slight degradation in "transients", but again it is virtually inaudible (to my ears, even using good monitoring equipment). The third thing is a loss of high frequencies - this becomes more noticeable as the amount that the speed is changed increases - for small changes (a few %) the difference is subtly noticeable with good monitoring equipment and good hearing. For larger changes (around 100%) the loss can be very obvious resulting in the sound becoming very dull.

(All three of these symptom essentially come down to a loss of bandwidth)

I believe that the difference between the Pitch/Tempo change and Speed Change is that the first two require "time stretching" (for which Audacity uses "SoundTouch") whereas for Change Speed all that is required is resampling. By default Audacity uses libresample, which is a high quality resampling library. If Audacity is built from the source code it can be built to use libsamplerate which is an even higher quality resampling library, but there are some licensing issues that prevent Audacity from being distributed with libsamplerate.

changing speed using Audacity (audible quality loss ?)

Reply #12 –
OK I did a little test, I took a nice crystal clear live recording of a track I'm very familiar with, I changed the speed of it back and forth 9 times (which is ludicrous!), and at least with my computer speakers I'm having trouble deciding whether I can really hear a difference or not, sometimes I think there is a very subtle difference, but I'm not even sure if my mind is just fooling me... So if that's the result I get from 9 changes I guess I don't really have to worry about 2 changes !

changing speed using Audacity (audible quality loss ?)

Reply #13 –
What do you mean by whether your mind is fooling you?  Do you not know what a double-blind test is?

http://www.hydrogenaudio.org/forums/index....showtopic=16295

changing speed using Audacity (audible quality loss ?)

Reply #14 –
Quote from: greynol on
What do you mean by whether your mind is fooling you?  Do you not know what a double-blind test is?

http://www.hydrogenaudio.org/forums/index....showtopic=16295


I've done similair tests....
I ended up mixing up the Audacity file (which has the speed changed 9 times) with the lossless original, so I didn't know which ones they were. I listened to one part of the song for 5 seconds (repeated listening a few times) and decided which I felt was the lower quality sounding one. I mixed them up 10 different times and listened to a different part of the song each of the 10 times (about 5 listens lasting 5-10secs to each part) and only 5/10 times I correctly chose the Audacity speed changed file as the file which sounded lower quality.

However when I did the same test with the lossless original and 128kbps CBR MP3 created with dBpoweramp the test only took me less than half as long (though I still did have to concentrate carefully), and I chose the MP3 as being the lower quality file 9/10 times.

So it seems Audacity speed change is doing a decent job, especially considering a file speed changed 9 times seems better quality than a 128kbps CBR MP3 of the original lossless file!

changing speed using Audacity (audible quality loss ?)

Reply #15 –
Those sound like very plausible results.  I would think it would be much easier and better controlled to use a dedicated ABX program or plug-in like the one found in foobar.

changing speed using Audacity (audible quality loss ?)

Reply #16 –
Quote from: greynol on
Those sound like very plausible results.  I would think it would be much easier and better controlled to use a dedicated ABX program or plug-in like the one found in foobar.


I'll check it out if you could please direct me in the right direction, I don't actually know what you're refering to.
Thanks

changing speed using Audacity (audible quality loss ?)

Reply #17 –
Quote from: Slipstreem on
Quote from: pdq on
With the amount of noise and distortion in tape, I would think that artifacts in digital resampling would be the least of your problems.

With an expensive Type IV tape recorded in Dolby C on a professional quality deck with Dolby HX-Pro, I wouldn't want to place any bets on that.


I've never seen a cassette copy of a CD that I couldn't ABX from the origional in a heartbeat.  Magnetic tape recording *always* imposes some pretty grim artifacts on the music that passes across it. Even 15 ips half track on the best studio machines can be ABX'd in the first generation.

ABX tests of tape generations

changing speed using Audacity (audible quality loss ?)

Reply #18 –
Quote from: bomber on
Quote from: greynol on
Those sound like very plausible results.  I would think it would be much easier and better controlled to use a dedicated ABX program or plug-in like the one found in foobar.
I'll check it out if you could please direct me in the right direction, I don't actually know what you're refering to.
Pinned topic in this very sub-forum:What is a blind ABX test ? (link to software at bottom of first post)

changing speed using Audacity (audible quality loss ?)

Reply #19 –
Quote from: Arnold B. Krueger on
I've never seen a cassette copy of a CD that I couldn't ABX from the origional in a heartbeat.  Magnetic tape recording *always* imposes some pretty grim artifacts on the music that passes across it.

Nobody made any mention of the source being a CD or of CD-quality.

Quote
Even 15 ips half track on the best studio machines can be ABX'd in the first generation.

ABX tests of tape generations

With a 57% success rate for original versus first copy, yes. The difference is hardly night-and-day. If it was then I'd expect to see it somewhere up in the 90s.

Having tracked down the technical spec for an Otari MTR-10-2 (I'll have to assume the exact model number as the ABX test linked to has much vital information missing) running at 15 IPS with a Scotch #226 tape, the cassette deck I'm specifically referring to (the TEAC V-1050) has it matched for SNR with the use of Dolby B and beaten by around 6dB with Dolby C, beaten slightly on low frequency response, and extends beyond 20kHz within the -2dB limit as specced for the Otari with any tape => a TDK SA.

Admittedly, wow and flutter is roughly a factor of two worse at 0.045% W.RMS versus 0.04% DIN 45507, but it's never been an audible issue for me with regular servicing.

I'm not saying that it's anywhere near perfect or even a match sonically, and I'm sure that the Otari would knock it into a cocked hat in terms of any type of distortion you'd care to measure, but the difference with this is hardly night-and-day either (to me anyway), with even some fairly subtle differences in mastering quality between source CDs still being clearly audible after being recorded to anything better than a sub-$1 TDK D-90. 

In an attempt to stop the hijack, I'll post on this subject no further, but I'll be happy to hear from you via PM if you wish to carry on this discussion and throw me a few pearls of wisdom. Your insight on all audio-related matters is always of interest to me.

...or maybe a kindly moderator could split our posts out to a new topic (possibly entitled "Are analogue tape recorders really that bad?") so that the discussion can continue in public.

changing speed using Audacity (audible quality loss ?)

Reply #20 –
Quote from: Arnold B. Krueger on
Quote from: Slipstreem on
Quote from: pdq on
With the amount of noise and distortion in tape, I would think that artifacts in digital resampling would be the least of your problems.

With an expensive Type IV tape recorded in Dolby C on a professional quality deck with Dolby HX-Pro, I wouldn't want to place any bets on that.


I've never seen a cassette copy of a CD that I couldn't ABX from the origional in a heartbeat.  Magnetic tape recording *always* imposes some pretty grim artifacts on the music that passes across it. Even 15 ips half track on the best studio machines can be ABX'd in the first generation.

ABX tests of tape generations
Oh, Arny... I couldn´t spot a difference. Back in 1993 when I got my first tape deck (Sony TCK-590). CD playback was done by a portable CD player from Technics. Using a Sony Type IV cassette (the white "Metal Master") I couldn´t hear which one was the CD and which one was the CD. The cassette deck could switch between the line input and the cassette it was playing - and if you weren´t looking and were switching fast enough you really wouldn´t know what was playing unless you were looking at the component again. Back then, I was very convinced that cassettes could sound (almost - leaving room for doubt) like the CD. But this cassette was so very expensive. Others weren´t that good so you could hear the difference.

To get back to the original question: why don´t you try the "Shift" function in Sound Forge? It works almost flawless. I did some testing on mulitple sine tones close to Nyquist and looked at them again after I did a speed up or slowed them down. With quality setting "4" there weren´t any errors, the function did not introduce distortions or some spurious elements. The only thing that didn´t work so well was the anti-aliasing filter if I was doing a speed up leading the particular sine to "tip over the edge" - but when you stay below nyquist it works pretty good. And you always can upsample...
marlene-d.blogspot.com

changing speed using Audacity (audible quality loss ?)

Reply #21 –
Quote from: Cavaille on
Quote from: Arnold B. Krueger on
Quote from: Slipstreem on
Quote from: pdq on
With the amount of noise and distortion in tape, I would think that artifacts in digital resampling would be the least of your problems.

With an expensive Type IV tape recorded in Dolby C on a professional quality deck with Dolby HX-Pro, I wouldn't want to place any bets on that.


I've never seen a cassette copy of a CD that I couldn't ABX from the origional in a heartbeat.  Magnetic tape recording *always* imposes some pretty grim artifacts on the music that passes across it. Even 15 ips half track on the best studio machines can be ABX'd in the first generation.

ABX tests of tape generations

To get back to the original question: why don´t you try the "Shift" function in Sound Forge? It works almost flawless. I did some testing on mulitple sine tones close to Nyquist and looked at them again after I did a speed up or slowed them down. With quality setting "4" there weren´t any errors, the function did not introduce distortions or some spurious elements. The only thing that didn´t work so well was the anti-aliasing filter if I was doing a speed up leading the particular sine to "tip over the edge" - but when you stay below nyquist it works pretty good. And you always can upsample...


Somebody much more knowledgable than myself has tried Sound Forge 8 vs Audacity and they believe Audacity seemed superior with the Speed Change function.