changing file speed with Audacity (audible quality losss ?)

By 'keeping you honest' did not mean I thought you were lying. No I am sure you are truthful and know what you are doing except I do think you are a bit confused and have miss quoted me terribly.

I did have fun with all of you because some 128 br music will sound identical to lossless because it lacks high tones that normally get removed with a 128 BR audio.

Except for being slow, Lame is an excellent encoder. It has the least defects and produces several formats to fulfill different requirements. However, at least to me, if you use Black & White film you really shouldn't complain about the lack of color. I tried to explain to you the difference between Lame CBR and Real Media, AAC and WMA CBR. Had you read and understood the link you posted I wouldn't have to explain this again. By the way all the members who joined in this thread are HA members. I know you posted this same query on HA as well.

I will try 1 more time...
Lame has 4 modes I only understand 3 of those modes. CBR, VRB and ABR. CBR compresses by truncating (removing the high pitched tones) to achieve a specific and uniform bit rate. VBR and ABR use psychoacoustic compression (PC) as does Real Media, AAC and WMA. This removes data you will not hear. It is lossy compression without actual quality loss. Because of how it works, PC is very spotty and can't be relied on to meet any specific compression. For this reason, they all employ truncation to deliver a more consistently compressed product. VBR truncates at a specific frequencies associated with a particular setting. See the table labeled Technical details of the recommended settings in your link. Out of all the compression schemes this is the only method that produces a predictable and uniform quality. At the highest setting the quality exceeds a 320 CRB mp3. (This is what I use.) Real Media, AAC and WMA truncate the highs to deliver a constant bit rate. The weakness of this scheme is the frequencies for compression are all over the board, for instance, if there is a moment of silence, PB will reduce the bits to 0 but the CBR will fill the space with empty values to achieve the specified bit rate. If there is a great deal of complexity, such as a violin solo that produces all sorts of resonance tones the truncation has to truncate much more than usual, cutting out much of the nuances. In a nut shell it cuts out the highs when it is most needed. VBR and ABR will leave the silence at 0 bits. ABR will use that reserve in spots were the PC is poor so ABR will not truncate as much of the highs as Real Media, AAC and WMA in other parts of the music where it is needed most so the quality will be slightly better even though the file size will be the same. Real Media, AAC and WMA produce commercial files to be sold or 'rented' that license agreements forced them to 128 BR. They comply and provide better quality than a 'normal' CBR. For these, the time index is always correct while VBRs and to a lesser degree ABRs do not work correctly. I feel that is a small price to pay for improved quality but that is left up to the user. HA claims Lame is virtually free of defects while HA claims Real Media, AAC and WMA have many know defects.

No encoder is perfect they all have pluses and minuses. It is up to the user to decide what issues are important then educate your self to know which provides the product with the qualities you desire.

I have wasted enough time on this thread.

 

I go back on my word for this brief statement K00ka understands me well. Yes I did mean ALL of you. There really are audios that will not degrade even at 128. I was being playful but reminding all of you not to be so free with sweeping statements.

Boomber, note there is a HUGE difference between being able to hear a tone and pick out that tone within music. As you approach the limit of your hearing the tone needs to be increasingly actually, logarithmically, louder so you can hear it. If you can hear it at 100 db that means you can hear it. The extremely high tones are usually harmonics of stringed instruments which are soft. Just because you can hear a tone at 100 db does not prove that you can hear that same tone in music because it is not at 100 db and is being masked by lower tones. The whole concept of PA is that you hear lower tones over higher tones. EX do you hear a violin play when a cannon goes off? No. The cannon masks the violin. The cannon data is minute compared with the violin.

Lastly, the reason mps do not go above 320 BR is humans can't hear what is being thrown out. 192 CBR and 160 VRB are thought to be transparent to adult humans. That is debatable but then you can always try it out yourself.

Thanks K00ka

 

I think I've been misunderstood.
Not only because I've been reading about MP3's for days when my original topic/question was about how well Audacity Speed Change function performs when exporting as lossless FLAC/WAVE.
Thanks to anyone who posted anyway......

 

I don't think you've

..We just have differences of opinion and preferences..I at least feel, it's been a good thread, thanks!..
Do drop by/post again, as I'm sure you'll have much to contribute..
And you're welcome Mez!!..
Cheers all.

 

Please don't go away mad. We have at bit more fun on this board the HA. More importantly, I hope you think about everything we discussed. Lossy audio is not to be ignored you learned a bit and may be better for it. What you picked up could enhance your understanding of the original project.

I think you know the answer to your original question yes. It will degrade the audio. Will you ever hear the difference? No Probably even if you twiddled it back and forth 100 times the best listener in the world will not be able to tell the difference. Why? There is an enormous gape between the detectable by electronic devices and detectable by the human hearing. That said, since you are a lossless man meaning to me you hold to trying to keep the detectable errors to a minimum shouldn't matter that you can't hear them. I suggest you keep a copy of your original then when you get the timing right, use the original and switch the timing only once that will preserve your data quality as best you can.

It has been a pleasure. Come again!

 

No worries, I'm not mad at all, never have been... ;-)
One of reasons I stated that I think I might have been misunderstood is because I was told not to take something so personally, and now I'm told not to go away mad, so I think I've been a little misunderstood, but seriously all is fine, there are no problems and haven't been, when you commented to keep honest, I knew you were joking, I never thought it was an accusation, my comment "why would I lie" wasn't supposed to be defensive which I think was the way you perceived it to be...

Even though I have wondered a little how I ended up absorbed in MP3's / 128kbs files for 2-3 days I still enjoyed reading everything everyone had to say, I wouldn't have done various blind tests of 128kbps files / MP3's using different Encoders/Settings etc. if I wasn't interested... I've learned a few things like the limits of my lossless/lossy radar !, I've been aware of it for years but never really pushed any boundaries with tests etc. It's actually got me very interested to see some statistics on what % of people can detect lossy from lossless at the different bitrates etc, something like that would be very intersting to see...
Again when I asked you if the Codec/Settings I switched to where the ones you were suggesting, I wasn't being defensive as it may have seemed, I was genuinely wanting to check with you that I had used the setting you were suggesting, because if you're telling me a specific codec does the job the best, then naturally that's the one I want to make sure I'm using !

Thanks again for all the info...
There's still a couple of things I asked which I hoped someone could please answer which I'm intersted to know eg. when using Lame VBR and I choose 130kbps, why does the output file say 146kbps ?
Another example with a different Codec Windows Media Audio 9.2 setting was 128kbps VBR but the output said 143kbps.
Is this normal or some kind of error ?, and are the files really boosted to 143kbps even though I chose 128 ?
Thanks again it been enjoying/interesting.

PS - When you say most likely Speed Change will not cause audible quality loss because you're working with lossless formats, it confuses me a little, for example on this very page Davexnet seems to say he tried Speed shift with "Sound Forge 8" with inferior results when compared with "Audacity", which tells me that just because you use lossless formats doesn't mean it's very unlikely (safe) that you won't hear any loss, because it seems like Davexnet noticed a difference with the first 2 programmes he compared, which says to me noticeable quality loss is extremely possible depending on which programme you use (unless the difference Dave is refering to was based on technical readings of some type and not actually based on what he could actually hear by ear ?).
However like I said, using Audacity I haven't so far been able to notice a difference when doing a blind test vs original lossless, I would be very interested to know if anyone with more expert ears than myself ever gets around to checking Audacity out.
Cheers

 

bomber07, when choosing a Lame VBR setting e.g -V5 the encoder will use more bits(higher) for complex parts in a track and less(lower bitrate) bits in simple parts..So it's normally for it to display say anything between say 120~150kbps..If you play the Lame VBR track you'll see the bitrate fluctuate up and down..
Back to Audacity, I have it installed on my PC's and lappy, but in all honesty I mostly just use it for recording streams and such, and don't do much if any editing..That said, I have no experience using any other editing SW, so not sure what if any differences there could be, especially when working solely with lossless> <lossless conversion/editing..
Technically there should not be any quality loss when working with LOSSLESS..If davexnet noticed "inferior" results using Soundforge, well who's to say he is wrong?..I'm sure the same goes for other Soundforge users that'll say the opposite..Again,'none of us share ears'..And we mustn't forget the 'Placebo effect'..

 

I am glad you didn't take me the wrong way.

At AD discussions often go astray and sometimes we are better for it. Mp3s have their place in the audio world, especially the HiFi ones. As the term suggests you can't hear the difference between them and the lossless original if built properly. Although I am a huge lossy proponent, I would never suggest using an mp3 for archival.

The 128 mp3 was probably a sanity check. We do get crazy persons asking questions or making claims. In most cases, even a person with diminished hearing can pick out even the better fidelity mp3s from HiFi. You passed that test with flying colors.

1) Testing would probably be an absolute can of worms! That is why if you ask that question they will tell you to try the testing your self. That is the best way but is very time consuming. One reason is the music picked for the test will skew the results you bumped into the problem with #3. Another problem is, adult hearing is not uniform. The best you will get is posted in the HA lame wiki. 160 VBR and 192 straight truncation are usually about transparent. Every piece is different every person is different.

2) Lame and Helix have the least defects Lame is virtually clean and Helix is nearly as good but is much faster. There is wide agreement VBR is produces the highest quality. The highest settings will give you the best quality. Slow analysis with the Lame encoder will slow the process even more but will improve the quality.

3) You hit the jackpot with this one! It is one thing to read about something and something else when you stumble over it. VBR is all about quality. When you pick a setting you are setting a quality standard and forget about the bits. The estimate is only that, an estimate. I would guess your sample is more complex and less easily compressed than the average audio.

I have personally experienced a range of something about 128 to over 300 bits for converting lossless to mp3 using Lame VBR V-0. Both of which I believed were mistakes and actually got worked up with the low one because I didn't catch it when I made it. I had to go back to the basement and root through my originals to find the original when I discovered 'the mistake' years later. I redid the disk with the same results and I thought the CD was bad. I got a lossless copy of a vinyl audio capture with the same results. The album was the Elton John album I mentioned before. The master lacked the expected highs. The other surprise was classical guitar solo. After you have been using VBR to compress music you will begin to understand the how VBR works. A good recording of a solo stringed instrument will compress a great deal less than that same instrument in a symphony. I know on the outside that does not make sense. There is so much more data in the symphony. However, in the symphony, there are louder and more easily heard lower tones that mask the complex, soft string tones so the result is greater compression.

4. Dave needs to answer that one. Maybe there is more chance of actual change than I believe there should be. What you are playing with may be less complex than what Dave is using. That is why you ought to implement the safest procedure you can. That way you do not screw up some audio capture after doing 25 tapes without a hear able change. I am going to sound like a broken record, converting lossless to VRB and observing the resulting average bit rate you will be able to guess better what music might be problematic due to its complexity.

Lastly, check this out when you get a chance unless you have already done so. This brief video will put perspective on what is detectable and what is hear able. This is really a must see for any serious listener. This provides an explanation why some persons claim they can hear a difference between HiFi lossy while science would indicate otherwise.
Acustic Myth Workshop

 

Double post

 

You mean I finally had my first cup of coffee, before Mez?..

 

No, I am on the East Coast, but my computer must have been on your time. I posted got distracted, then when I looked at the computer I 'realized' that I hadn't posted it so I posted it again. I think that was my first double post ever. Even though I am a night person and loath the morning. For most of the jobs in my long carrier, I have needed to get up early. I am a programmer. Management usually allows programmers to have great flex hrs so, if they need you to be somewhere at a terrible time you comply with a smile if you are smart. For the last 20 yrs I have had long commutes. I live out side DC. For my last 2 jobs that totaled over 12 yrs, if I left at 8 AM I might get in at 10. If I left at 5:30 I would get in at 6:00 +/- a few minutes. I could also leave at 3 so I could get home at 3:45. If I left at 8 I would have spent 4-5 hrs on the road every day. The time cost and even the gas cost were so great, it was well worth leaving so early. Now, I usually get up early.

 

I did a test comparing SF8 and Audacity to the original.

Now that I listen to it I hear a slight diminishing of the sound in each,
SF8 appears to add some artifacts while the Audacity stretch not as much so,
but it loses some of the crispness of the original.

It's generally accepted that this operation will degrade the sound slightly.
Whether you can hear it and whether you can live with it is down to the individual.

I've often done it on the sound track of downloaded avi files where I want to do
a PAL to NTSC conversion; apply a stretch of 104.27% to slow down the audio.
But in that case a loss of fidelity doesn't bother me. I just want to watch the
file and be done with it.

In this test the original file was 30.1 seconds (approx). In SF8 and Audacity
I stretched to 32.4, then 28.9, then back to original size.

http://www.mediafire.com/download.php?q1lfqkrcenvnasn

 

Dave, I will take your word for it. We agree it has to alter the quality. Since I am unwilling to play with it my self, I am content with your statement. You may have sharper ears or better play-back equipment than bomber or maybe you picked a nasty piece of audio while bomber was converting something that needed work. If I did the test, I would pick something that didn't compress below 300 BR in a VBR compression. I would figure anything so intricate would expose flaws.

You could hear the diff with only 2 changes? Again, my advice was keep an original, play with a copy then make only one change to the original to play it safe. If it were me, I would leave it alone since I liked listening to the tape. I tend to be a bit sloppy with my mp3s but not archives. However, I don't even play with mp3s. I prefer the original sound since it brings back memories.

 

Mez, I do not have a high-end audio PC.
I'm using basic onboard sound and $20 speakers.
But the SF8 process does cause some artifacts that are not in the original
and can quite easily be heard. IN fact, in the SF8 dialog, it actually warns
in some of the modes that artifacts may occur ! (see below)

I did 3 changes. I lengthened it, shortened it, then back to original size.

The 10mb file I linked to includes the original and the SF8 and Audacity files.



Uploaded with ImageShack.us

 

Holy crap! A picture is worth a 1,00 words. Even at that low res pic you can instantly spot differences with out any blow up. My question is what are you comparing? You talk about 3 files. Original, SF and Audacity. Which are those two?

Well it doesn't take much of an ear or equipment to hear some artifacts. I have heard some that startle you.

Now I know why a friend of mine spent a grand on an audio editor.

 

I showed that as an example, it's not the file I used in the test,
I just wanted to show the dialog box.

The song in the file I posted is a Radiohead track (30 second clip).

What you see in the screen shot is the left and right channel of a stereo track.

 

Thanks for the additional info Mez, interesteing reading again, will check out the link.
Davexnet, thanks for running through those Speed Changes and thanks for the 3 audio samples, I'll definately put my ears to the test when I have some time this afternoon.
EDIT - I just had a spare 3-5 minutes, and listened to Davexnet's 3 samples, compared to the original file the Audacity file sounded the same to me (but I really didn't have much time at all to listen), however to my ears the Sound Forge 8 file is obviously different to the original, I can hear it quite clearly in the sound of the hi-hat within the first 5 seconds... (why is it that I can best hear these differences when listening to the hi-hat ?, I hear it as clearer&louder vs less clear&softer). Probably not a very technical way of hearing it! hahaha

 

I guessed that you would link to the experiment image.

So what is seen here is way different because it isn't the same music but 2 channels.

You will have to help me out, I do not know how to view the image of the experiment. If I click on the link I get an upload screen, no image. If I open the link I can only see the same image and a windows update screen if I click on the arrow icon.

 

Are you trying to download the audio samples?
See the mediafire link above.

 

Well, Dave you got me so curious I tried it myself. I only slowed a particularly nasty piece 4% with Audacity. I am with Bomber; I couldn't tell the difference between the 2. My good speakers are starting to lose the glue. To re-glue they are only $350 a piece but if I blow them out I will never afford to get them fixed. So I converted the the lossless files to a maxed VBR and played them on my ipod. My buds are extremely high fidelity, better than my ears. However, my ears are much better than even $100 3 piece computer speaker set, which are really junk.


Maybe starting with avi files was a bad idea for an experiment that really required lossless audio. Anything less will magnify the degradation process. Bomber and I used lossless. I don't have any wave files but I have plenty of flacs.