Recently I found that I had two copies of a several tracks in my collection, one copy that I'd transferred from vinyl and another copy I'd acquired as mp3's (typically of dubious origins though probably sourced from CD). Making a few listening tests I noticed immediately that my vinyl transfers all sounded a little slow. Making some waveform measurements I estimated the speed error to be about 2% to 2.5% (too low) for my vinyl transfers.

The first thing I was wondering is if most people would consider this a serious problem or not? I was thinking about correcting the speed using software. I think audacity for example can speed up or slow down the audio so if I've got lossless copies it would seem fesible to correct it this way. Does this seem an ok way to do it?

The other question is : does this mean my turntable is crap? I just measured the speed, with only the platter spinning and no record or needle drag, and it was only going at about 32.7 RPM. There is no speed adjustment, at least no external one that I can see. It's a JVC belt drive turntable (I don't have it here with me at the moment so cant give more details). Just wondering if there's ususally any way that you can open them up and adjust the speed?

Does anyone else here tweak the speed of their vinyl rips in a wave editor. If so what procedure do you use?

Thanks.

I guess it depends on whether it's going to bother you every time you listen to those files that they are too slow. (It would bother me.) If they are lossless, I don't see how any real harm could come from reencoding them.

Foobar2000 with SoundTouch DSP component can batch reencode files to a different speed. I've used it to adjust bootlegs made from off-speed cassettes.

As for the turntable, a new belt may help. Some tables do have an internal adjustment, but usually it's not well marked. You may need the service manual to find it. Otherwise, if you primarily use it for ripping vinyl (and not listening directly), the suggestion above will fix your files.

Well the speed issue didn't really bother me until I made an A/B comparision with a file ripped from CD and then it was obvious. Anyway now that I know about it then it'll probably irk me, so I think I will correct them.

Hold that thought. ;-)

I don't know how anybody with perfect pitch ever tolerated listening to LPs unless they had a turntable with some kind of continuously-variable speed control.

Yeah, 2% speed tolerance is very poor. Sooner or later I'd think it would be likely that it would get on your nerves.

For best quality I would suggest using Audacity to resample the audio because of the relatively high quality resamplers available. But if the speed changes continuously - quite common with most belt drives - your only options are to either repair the table or get a new one.

Really? http://www.hydrogenaudio.org/forums/index....st&p=562732

Well, yeah, but compared to SoundTouch?

SoX seems to have one of the best resamplers (and hence speed changers) around -- you can use notepad to edit its example batch file with "speed 1.02" (remove the rate command) then drag-n-drop the files to be converted.

-bandpass

Yes continuous speed variations of that magnitude would be dreadful, it's nearly half a semitone. Fortunately however it seems to be a fairly constant speed error rather than fluctuations. I think that was the point that Arnold was making, that you shouldn't really notice a constant small offset unless you had perfect pitch hearing.

As for resampling, I've always been very pleased with the quality of foobars resampler so I'd happily use that. I know this sounds a bit "dirty" was just thinking, even if I just resampled to a 2% lower sample rate and then just rewrite to wave header back to the original rate (eg 44100) then that would achieve the speed change right?

I should point out that I'm not really super quality conscious in this area. I don't listen to vinyl very often but I like to keep a working turntable setup for the odd occasion when I do. These days if I find something on vinyl that I want to listen to I usually also record it to computer for the convenience of having it with my digital collection. My transfers are not the highest quality (most of my old vinyl is not in that greater condition anyway) but I still find it ok to listen to (I grew up with vinyl so you kind of get used to the imperfections). I'd like to post a few samples but I'm away for the weekend and don't have anything here. I'll post some samples on Monday and would really appreciate any opinions on the relative quality level.

Thanks bandpass, yeah that Sox resampler looks pretty good too.

Some tips for the turn-table: take the belt off, clean it in soap and warm water, this alone may fix the speed issue. If not, then leave the belt off for a couple of weeks, it may well recover. If so, when you've ripped a disc, take the belt off again until you're ready for another rip.

-bandpass

There are a number of ways to *correct* the speed of your turntable. Depends on what sort of technology they use in the motor.

If the motor is a straight up AC motor, then you can make up a variable frequency power supply for it, out of a cheap audio signal generator, a small power amplifier, and a low-voltage power transformer hooked up backwards to step-the voltage of your power amp up to 120 vac.

If this is a typical belt drive turntable, you can decrease the speed of the turntable by increasing the diameter of the driven pulley. One simple way to do this is to wrap the required number of layers of thin adhesive tape. If you get a nice tight butt of the ends of the tape, there won't be a momentary speed change at the point where the ends of the tape meet.

Another approach is to spray paint the driven pulley.

If the driven pulley is 5 inches in diameter, you will need to add 1/50 th of the diameter or 1/10 th of an inch to make the platter spin 2% slower. That's a lot of paint or not a lot of tape.

Another approach that may work with some turntables is to add an electrodynamic brake by putting a very strong mangnet near to, but not touching the aluminum platter. Adjust the distance to get the desired braking effect.

I've used the tape technique, but I only needed a fraction of a percent adjustment.

Something to consider: you never noticed until you compared it to a reference, so really it is intrinsically NOT noticeable.

Ok thanks for the info Arnold. I don't think I would be enthusiastic enough to build a variable frequency drive for this old turntable but the tape trick sounds like a pretty easy idea to try.

I'm thinking about putting a few layers of tape onto the driving pully (the small one). I'll measure the diamater and work out how many layers of tape will be need to increase it by 2%. Thanks for the idea.



Yeah that is definitely something I've considered. I'll post some samples tomorrow and see what people think of them.

Ok here's some samples if anyone would care to listen and comment.

The first one is "My Best Friends Girl" by "The Cars". I've included a short sample of the vinyl and the mp3 from CD for comparison. I made this vinyl transfer a few years ago and I haven't corrected the speed. I can't really remember how much post processing (click/pop removal and denoise) that I did on this one but I'm sure there was some (maybe even too much). For comparison I've included an mp3 sample which I presume is from CD, though I have to admit that I downloaded this mp3 and have no real information about it's origins.

The second sample is one I just transfered from vinyl today. The track is "Breakfast at Sweethearts" by "Cold Chisel" (they're an Aussie band from the late 70's / early 80's). I corrected the speed here, though it only needed just under 1.5% which is an improvement since I cleaned the belt and lightly oiled the spindle to make it spin as freely as possible. I also declicked and denoised this one but only fairly "lightly". For comparison I've included a sample from the CD as well (yes I do actually own a copy of the CD for this one )

I've zipped all the samples into a single download. The files size is 5MB and contains 4 samples. All samples are vorbis at about 180kbps, except for the "cars" mp3 that I only have as an mp3. For this one I just cut a sample using mp3directcut.

http://rapidshare.com/files/166951803/vinyl_samples.zip.html

Thanks.

It's interesting to read this tape solution. When I was a teenager, I had one of those cassette/turntable/radio things that were popular. The turntable played fast to the point that the pitch was raised a half step. Since I was learning to play guitar, it drove me nuts.

One day I decided to see what I could do about it. I took the platter off and saw no adjustment. What to do... I figured I either needed to add to the rubber drive wheel or reduced the metal spindle out of the motor. Putting tape on the rubber wheel just didn't work. So, I took a piece of emory paper and held it against the metal spindle while is spun. Being careful not to take too much off at once and doing repeated measurements using my guitar tuned to concert pitch, I got it to work. I'm sure it wasn't exact, but it was good enough for my ears.

This experience caused me to only consider turntables with a strobe and adjustable pitch when I went shopping for my real turntable (which I still have today).

The two samples are *quite* different. The first note of the the LP file is centered at 169 Hz, while the first note of the MP3 is centered at 175 Hz. That's about a 3.5% difference. There are very strong differenced in dynamics as well.



The difference here is 60 Hz for the CD sample and 64 Hz for the LP, a difference of about 8%!

BTW you notice that the LP file in the Cold Chisel set is sampled at 48 KHz and the other is sampled at 44 KHz?

If I correct that, the files play at the same speed.

The second samples are very interesting. What are the components of your vinyl transcription setup?

Usually, there is no internal speed adjustment. And usually, something like a worn belt will result in "WOW" (slow wobbly speed variations). Something like that is unlikely to result in a constant 32.7 RPM. (So, it's probably not a "simple" repair, but I'd give it a try anyway.)

A good quality turntable can have a synchronous motor (that locks to the line frequency). In that case it will not have any adjustment. The line frequency is very accurate in developed countries... Or, it can have a font-panel speed adjustment with a built-in strobe. The strobe is usually run from the line frequency... Or, it can have a quartz crystal oscillator (like a soundcard or quartz watch).

Thanks for taking the time to listen to them Arnold.

Yes the transfer I made yesterday was sampled at 48kHz and all the others were 44.1kHz, but the headers are correct so it should still play at the correct speed.

The choice of sample rate wasn't meant to unfairly advantage anything, it's just that the soundcard I was using works better at 48k than 44.1k (in RMAA loopback tests) so I recorded it at 48k. I usually resample from 48k to 44.1 (mostly just for compatibility if I want to burn to CDA) but yesterday I was a bit pressed for time so I left it at 48k. It shouldn't make much difference since the lossy codec will have capped the bandwidth to under 20k anyway.

Regarding your question about my setup for yesterdays transfer.

Turntable : JVC JL-A15 (Victor Company of Japan)
Amplifier : Yamaha AX-396 (Using phono-in and line out)
Soundcard : Realtek HD Audio (onboard yikes )
PostProcessing : "Pristine Sounds 2000" software for de-click and de-noise.

Usually when recording I use my older computer which at least has a dedicated (PCI) soundcard. It's only an SB-Live but in loopback tests it performs quite well. Yesterday I was so pressed for time that I used my "work" computer which only has the realtek onboard audio. Actually onboard audio seems to have come a long way in the past few years and this one is really not all that bad. In the past I've seen onboard audio (particularly on cheap motherboards) that was absolutely horrendous, but I've done loopback tests on this current one and surprisingly they were quite good.

Hi Doug. The more I look at this the more I suspect that a new drive belt will help with the speed. Since just cleaning the belt the speed has gone from -2.5% before to now only -1.4%. Also when I try to hold the motor spindle (with the belt off) the speed characteristics seem quite stiff with plenty of torque. I havn't worked on turntables before so I don't have anything for comparison, but the belt does seem faily loose and the platter can be stalled with only minimal torque (the motor still spins but the belt just slips). When I get time I'll look around and see if I can source a new belt locally.

BTW this turntable was mothballed for nearly a decade in the 90's. Mothballed isn't really the correct term because that implies it was stored properly for preseveration but unforunately it was more like "junked and left for dead" in a shed. When I revived it a few years ago it was looking pretty stiff and sore. There was some corrosion of the aluminium parts and I had to run it for quite a while just to "free it up" so that it would spin properly. (Actually now I think about it I recall that at first it wouldn't even turn - crikey - I had to spin it by hand to get it started, but once it was started I just let it sit there spining for a long time and then it was "ok"). Anyway the only other thing I did at the time was to get it a new needle, now I guess its time for a new belt too.

When I bought a new turntable (a "good" one) I found it ran perfectly to speed. The few recordings I owned on vinyl and CD matched perfectly.

Unfortunately, I'd been less expensive turntables all my life, and all of these ran fast. FWIW, IME, most less expensive tape decks run fast too.

So I was used to all my records playing slightly too fast - the correct speed sounded too slow.

My solution? Wrap some tape around the motor pulley!


(About a year later I took it off, and have "suffered" correct pitch/speed ever since - it's fine for "new-to-me" records, but some well listened to "old-to-me" records still sound wrong to my ears!)

Cheers,
David.

I had a bit of fun comparing the two Chsiel samples. They turn out to be very close in my casual listening tests. I could ABX them pretty easily, though. I've seen far, far worse, surprisingly made with far more reputation-intensive hardware, such as Linn.

Thanks Arnold, your comments gives me some confidence that I'm making a reasonable job of it, at least for casual listening which is all I really need.


Arh the opposite problem to what I've got. Personally I don't think I'd notice a small speed difference unless I played the two different speed copies of the track one immediately after the other.

Actually it was kind of funny how I discovered this problem by accident (even though you'd think it was glaringly obvious). I has just installed the latest foobar and it has a new media library feature which apparently had found some old vinyl transfers I'd forgotten about, and it put them in the playlist (though I didn't realize this at the time). Then just by pure co-incidence while playing about 2000 tracks in "shuffle track" mode it played the CD-rip versions of "The Cars - Best Friends Girl" followed immediately by my vinyl transfer version. Suddenly I couldn't believe the difference, it sounded like the producer had just given the band members a bunch of marihuana and Rohypnol, then put them in a padded room and got them to record it again. It certainly sounded laid-back compared to the original.

Hehe.. This reminds me of a particular song that I originally heard at 45rpm's instead of 33 (it was mixed with another), and when I got one Mp3 of it some years later, i used to play with +12% speedup with a pitch change plugin in Winamp XD

(Yes, this made a very big difference)

Hi,

I have done some very precise work at audio recovery. I suppose I am more exacting because I have background working in a radio studio.

Over the past ten years I spent a great deal of time recovering audio which in most cases originated from vinyl. Sometimes the route was in-direct (as the vinyl was no longer around and it had been transfered to analog audio tape.) That obviously added another area of recovery... but in regards to adjusting speed, this certainly became an issue as often the original recording deck did not match the current playback unit.

Concentrating on speed, I am fortunately a musiican so I got in the habit of taking out a reliable pitch source, in this a case an accordion. It's easy to hold a steady note. Utilizing software that can increase or lower the pitch of audio which has been transfered to digital medium one can match slightly off-tones to see just how far off the speed of your recording became through it's various stages of dubbing. Of course one does not need an accordion, and I would not recommend that you try this with an acoustical instrument (they are not pitch reliable), as an accordion is actually a metal reed instrument. Tuning fork, electric tuners, I assume will be less pricey and available to the average person.

There is an exception to this. It is obvious that there have been some moments in which studio engineers have decided for one reason or other to manipulate the recorded pitch which is mastered in a recording. Always, there are exceptions.

It could get worse. I have had a few precious recordings in which the audio speed either varied constantly due to an off-center support hole in an analog disk. I have been dreaming of an inexpensive software solution to that.

And of course, perhaps you grew up with these recordings sounding precisely off-key as they are now that you have transferred them. I personally found listening to RCA's 8-track version of "Fat Angel" by the Jefferson Airplane", which brings to memory the Chipmunks and Mickey Mouse, a bit amusing and thereby fond to my heart.

My bottom of the range 80s table runs a tad slow at 33 1/3 and pretty much spot on at 45.

Heh - I'm reminded of a Schoenberg record I have that skipped/locked grooves in precisely a position that made it seem like it was a deliberate part of the music. Schoenberg meets John Cage, I tell you!

Bear in mind there are different kinds of resampling. Some will speed up the song but not change the pitch (not what you want). Some will change the speed and pitch the same amount... essentially saying the song was recorded at 43 khz in stead of 44.1 for example.

Many turntables (adjustable ones anyway) have a pattern with lines of dots on the edge of the platter and a neon bulb shining on it strobing at 50 or 60 Hz (depending on your country). WHen the speed is right, the dots for that speed appear to stand still.

For other turntables you could get a paper disk with a similar pattern to put on top of the platter. This one was on wikipedia. Maybe you can find one online for 60 Hz that you can print out.