[neil]

Recently, I encoded Coldplay: Parachutes using Media Jukebox. I encoded the entire album, and I decided to listen to the CD on a jambox (so my wife could hear), but I wanted to have my Grados going too (for better quality) and as I sycned them up, repeatedly, the MP3 would fall behind.. and it was quite noticable just after a couple measures of the song.

I used LameMP3 VBR (highest quality). I'm going to try GoGo MP3 encoding method to see if that does the same thing. Could it have been my CD player on my notebook? Could it be my jambox is playing fast? It's really weird.

And if you're wondering if I'm imagining this, I'm not. I'm a drummer with some decent rhythm and it's definitely losing tempo -- and what's really weird is that the TUNE doesn't seem to be off.. but I'll listen more closely.. in fact.. just did.. track six this time, and indeed, it's detuned slightly.

Yikes. Lets hope this other encoding doesn't do the same thing.

 

[Cousin Eddie]

I wouldn't think that the mp3 encoder would be inserting delays. I think that other things are probably just interrupting/hogging resources.

Personal computers were never really meant to be real-time devices. The architecture of Intel PC's is interrupt driven, which is not an inherently good design for doing real-time applications with multitasking. If you think about it, there's really not much that needs to be done in perfect real-time with computers. It doesn't really matter if the cursor stops a millisecond while you're typing something in Word. Even in games, the slight delays aren't a problem because things are always being loaded off disk and you have to wait anyway.

That's not to say that you can't do real-time things with personal computers. You just have to optimize everything so that you don't have lots of stuff running in the background and constantly generating interrupts and taking processing time. Real Jukebox is one program that really hogs resources. But lots of programs running in the background do it too. Even the program that runs with my keyboard does it. And it's not the processing time that they take that is so bad, it's that they are constantly causing stalls. I use my computer to play DVD's a lot. This is what first motivated me to change what program's my computer was running in the background. Without stopping all those programs from running, DVD playback was choppy and ugly, but now it's perfectly smooth.

 

[neil]

It's not to say that the MP3 playback is choppy or ugly. It's actually very smooth and beautiful; however, it is playing at a very slightly slower speed. This leads me to believe that the problem has occurred during the encoding process, not during the playback. This can probably be proven by just copying the same file to a standalone MP3 player -- at which time I'm certain that I'll come to the same conclusion that its the encoding process, or possibily the inherent detriment to using MP3s regardless of encoding method (?).

Let me get more scientific in my tests -- and I'll report results. However, if anyone has any solid evidence (i.e. articles, first hand experience) on this issue -- please share.

 

[rickcr42]

Could be the CDROM drive in your computer having power supply issues : i.e not enough drive from the computer power supply.This could happen if you multi-task while ripping.If you have any DIY skills you might look into beefing up the psu caps leading to the CDROM motor drive.Even better would be a dedicated supply feed,regulator ,inductor,filter caps-all from the main buss

 

[neil]

Recording #1 - Test A
Encoded with: Dell notebook CD player, digital playback, lame mp3 vbr
CD playback test with: Dell notebook CD player (same we encoded with)
Comparison: MP3 speed is slower, almost un-noticable. Takes a 70 BPM song about 4 minutes until you notice it's
a few milliseconds apart.

Recording #1 - Test B
Encoded with: same as #1 above
CD playback test with: Philips/Magnavox jam-box
Comparison: MP3 speed is noticbly slower. With a 150-180 BPM song, it takes only a couple measures to actually
hear the speed differential.

Recording #2 - Test C
Encoded with: Dell notebook CD player, analog playback, lame mp3 vbr
CD playback test with: Dell notebook CD player (same we encoded with)
Comparison: MP3 speed is faster, noticable. With a 150-180 BPM song, it takes about ten measures to hear the speed differential.

Recording #2 - Test D
Encoded with: Same as #2 above
CD playback test with: Philips/Maganavox jam-box
Comaprison: MP3 speed is noticably slower. With a 150-180 BPM song, also takes only a couple measure to actuall hear the speed differential.

Recording #3 - Test E
Encoded with: Dell notebook CD player, digital playback, Ogg Vorbis 350K (highest)
CD playback test: Dell notebook CD player (same we encoded with)
Comparison: Ogg Vorbis is slower, almost un-noticable.

---

Results

My initial observations are that my Dell notebook CD player and the Philips/Magnavox do not play at the same speed. It seems that either my Dell is playing them back too slow, or my Philips is playing back to fast.

Another observation is that the digital playback during the encoding has a much tighter match, in terms of tempo, to the analog playback encoded MP3. Sound quality have yet to really compare between the two types of encodings.

My last observation is that the Lame MP3 and the Ogg Vorbis encoding methods did encode files at close to identical speeds. In recording #1 and recording #3, everything was the same, with exception to the encoding method.

Initial questions to Head-Fi community: Have any of you noticed any speed differences between your CD sources? I'm going to do some stop watch testing to see how much time it takes to get from point-to-point in a song. I'll test a couple Panasonic's I have at the office, a couple Sony's, a Radioshack, and ultimately against Jude's higher quality sources.

 

[rickcr42]

I do not think this speed issue has come up anywhere else,at least I have not heard of it before.But you may be onto something with even bigger implications,such as :

1) CD/CDROM transports and ARE there speeddifferences ?
2) What about MP3 encoders ? Do they change the tempo ?
3) Is this hardware or software related ?
4) I wonder how many transports are considered "lively" due to speed increases

You raise a lot of questions Neil , and further invectigation is needed here.Others,myself included,need to perform similiar tests and compare notes here.Maybe then we can get the big picture of exactly What is going on or is this just an isolated case.

 

[gorgon_123]

wow.. very interesting neil. i'll be tuned in for further info.

 

[gorgon_123]

Neil,

hope you read this. Anyway...

I emailed the hardware people at http://www.dmusic.com and got this response:

"ahhh after reading the first post of the thread I noticed that the mp3 was
encoded in Variable Bit Rate (VBR) which lowers and raises the bit rate
according to the output rather than a constant bitrate. When things vary
like a VBR does there is bound to be some effect. It would take a lot better
code writing to get vbr timecode to run better. CBR time is nearly exact.
Sometimes silences may be added to the front or the back of the track
depending on the encoder.

Joe Hui
Editor
http://hardware.dmusic.com"

Hope this helps!

 

[gorgon_123]

more information if anyone is still interested: a colleague of my uncle's found that his MP3s were playing faster than the original recordings. It turned out that he had a problem with his soundcard drivers. Once that was corrected the MP3s played back fine.

 

[neil]

I wonder if it was his soundcard drivers causing it to record at a slower speed, or if they were playing back at a faster speed. Because, surely, if you record and playback on a sound card it isn't quite the same as an analog recording devices such as a tape-recorder which uses the same motor during playback/recording.

The great mysteries of encoding. I still haven't had much time to do anything in this testing arena.. I'm afraid that with facts like what your uncle presented, and from what I've found in my tests, we'll come up with an explosive number of scenarios with different outcomes.

The test I'm still interested in is verifying the pitch/speed of CD players. Jude and I are going to borrow a co-worker's analog tuner and compare the output of portables and home players to see if they all play a note the exact same way (we have some test/diagnostic CDs).

Hey Jude, I'd like you to do that testing and have the report posted on here first thing in the morning.. and darn it, this coffee is stale!