Skip to main content

Yesterday I came up with this idea of taking digital audio, expand it to twice the duration (equals half-speed), still keeping it at 16bit-44KHz and record this halfspeed material onto a Revox B77-mkII at 3,75"/sec.

Then I played it at 7½" and it was quite a surprise to hear, that high frequencies was brighter, clearer and more crispy, than if I just record it straight away from source to 7½".

I have this theory, that tape handles 5 KHz better at 3,75" than 10 KHz at 7½", and so on. 10 KHz@3,75" versus 20KHz@7½"

What happens and why does it sound better ?

Comments

hueseph Tue, 11/27/2007 - 08:47

DK6400Brian wrote: Yesterday I came up with this idea of taking digital audio, expand it to twice the duration (equals half-speed), still keeping it at 16bit-44KHz and record this halfspeed material onto a Revox B77-mkII at 3,75"/sec.

For one thing you're doubling the frequency that you recorded. So yes, you are reproducing a higher frequency than you originally recorded because you've doubled the speed of the tape on playback. There is no more on the tape than there was when you recorded it.

The thing with recording at higher ips is that you get more information on tape. Kind of like recording a 96k instead of 44.1. So, ideally recording at a higher ips/samplerate will produce a more "true" replication than lesser samplerates/ips.

Boswell Tue, 11/27/2007 - 09:04

DK6400Brian wrote: Yesterday I came up with this idea of taking digital audio, expand it to twice the duration (equals half-speed), still keeping it at 16bit-44KHz and record this halfspeed material onto a Revox B77-mkII at 3,75"/sec.

Then I played it at 7½" and it was quite a surprise to hear, that high frequencies was brighter, clearer and more crispy, than if I just record it straight away from source to 7½".

I have this theory, that tape handles 5 KHz better at 3,75" than 10 KHz at 7½", and so on. 10 KHz@3,75" versus 20KHz@7½"

What happens and why does it sound better ?

It won't necessarily sound "better", it's just wrongly equalized. When you switch the tape speed, the equalization components in the record and replay amplifiers are also switched, so unlike digital reproduction, recording to tape at one speed and replaying at another results in a non-flat frequency response (as well as the signal frequency shifts). Theoretically, you could pre-equalize the digital signal to compensate for this, but it would take a bit of fiddling around stretching and subtracting the standard tape equalization curves to find out how to set the digital EQ.

There may be some additional saturation effects in the replay head because the induced signal amplitude is tape velocity dependent. This is not like unpleasant digital saturation, but more like the effects of a nice compressor. Transfer via tape is sometimes used to impart to a track or to a final mix a quality that is difficult to achieve in the digital domain.

anonymous Tue, 11/27/2007 - 09:38

Yes, that's true.

EQ is different at the different speeds.
Perhaps that's the main cause of the slight lift in the high area.
Like the days, when recording a CrO2 cassettetape and playing it at "normal" position ?

I was actually aiming at getting lower THD with this procedure, since I think to remember something about relatively poor performances from tape in the 10-20 KHz area. THD sets in pretty soon at level -12 dB.

I thought and hoped, that this could be a benificial "workaround", recording halfspeed material 10Hz-10KHz at halfspeed, playing it back at full speed, without having the taperecorder to deal with the limitations from the magnetic layer in the very high frequency area.
Apart from the EQ fact, haven't I found something here in any case ?

With regards to saturation, I was aware of the fact, that halfspeed material contains much more energy in the lower parts of the frequency scale.
That's why I carefully made sure, not to exceed an input level of more than 0dB on the VU meters.
Easy to handle with digital material.

RemyRAD Wed, 11/28/2007 - 01:16

I understand what you're trying to accomplish DK6400Brian. It's a noble attempt.

Because all, most all, analog type recording systems include some kind of preemphasis of frequency response for recording and deemphasis of frequency response during playback, reciprocity must be maintained for flat response. When you start playing with the speed phase anomalies happen and because equalization is not actually boosting or cutting anything, it is a phase manipulation, which means that low frequencies and high frequencies are traveling at slightly different speeds. This means any consistency is out the window when it comes to equalization networks and standard mathematical representations.

In years gone by, similar scenarios were utilized for a different perception in analog disk mastering. Some folks found that playing the master tape at half its speed and recording the lacquer master at 16 2/3 rpm would create an 33 1/3 rpm album with a different textural perception. Some of this was based on the fact that equalization networks could not be precisely described for this application. So, for those that felt that their product would sound perceptually better through half speed mastering and could afford it, would have it done. Instead of half speed mastering, I elected to go with DMM or, direct metal mastering. This was the same old process done the same way as with lacquer, at real-time speeds but not using lacquer. Copper was used instead and with that, came a change in the perceived texture and high frequency transient response.

So keep up your experimentation. You never know what you might trip over?

For instance, I'm transferring some 78 rpm historic records. I had this hair up my ass about half speed transferring. I thought the inertia of the needle would be more precise. Well, it wasn't exactly half speed but at 45 rpm. I would still record at 16-bit, 44.1kHz. Then, I would have to change the sampling rate to 76kHz to put it back to the proper pitch and speed. Now this mathematically does not quite work. It doesn't quite work because 78 rpm records were never actually at 78 rpm. This was due to older motor technologies so 78 rpm records were generally between 74 to 82 rpm average. This 76kHz of mine was accomplished by knowing what key the music was to be in. Then, I would have to downconvert/sample rate converter to 44.1kHz again for CD release. In a sense, that made the original transfer more high-definition since it required playback at 76kHz. Well, after numerous attempts to improve transient response, which worked well, I also doubled the frequency of the turntables inherent rumble frequency which before this process, was largely in audible. Now it was too audible so, I decided the benefits did not outweigh the detriments. So I am transferring at 78 rpm and still having to adjust pitch and sample rate, to put the recording into the proper key of music it was written to be in. And then still having to do a sample rate conversion to 44.1kHz again.

I don't play ball but I'm a good pitcher
Ms. Remy Ann David

JoeH Wed, 11/28/2007 - 06:48

I feel your pain with the 78 RPM transfers, Reemy. I too get a fair amount of material to transfer, LPs, 78's, Cassettes, all kinds of analog tape. It's amazing what's STILL out there, looking to be put onto a CD.

Since 99% of the material we get is musical stuff, we're always asking and checking about the original key it was performed in. (Sometimes if it's a work not very well known, we even have to ask the composer - or the composers survivors - to find out what key the piece was in, so we can START with the correct speed.)

Is it possible you can correct the TT speed BEFORE you do the transfers, so that it's already pitch corrected, etc., while it's going into your A/D? Just one less step in the digital domain, if you can help it.

RemyRAD Tue, 12/04/2007 - 00:46

You're welcome Brian, I hope my blather helped.

Yes, Joe, I can tweak during the transfer. Since I'm not going to do the half speed thing, that won't be a problem.

One of the things I'm rather pondering over lately is dealing with reel to reel archive transfers. Some recordings were recorded with old-fashioned machines without constant tension/torque and the speed drift that occurs. Other recordings were made with tension control. I loved the constant tension machines, since you could make musical edits anywhere within the tape. Folks just don't know what they're missing.

I'm not sure I like IE 7?
Ms. Remy Ann David

BRH Sun, 12/30/2007 - 17:53

I must have missed this thread.
Why not play back the disc at the correct speed in the first place, with a varible speed turntable, and forget about this sample rate adjustment/conversion. Just tune it on playback, you gotta listen to it anyway, so might as well be in the right key and in tune.

RemyRAD Mon, 12/31/2007 - 14:08

The reason is I was looking at the fact that there is, because of the higher rotational speed, the inability of the stylus and tone arm to sometimes track & respond quickly enough, since velocities are higher. By playing it slower, the inertia is lowered and response time of the stylus cantilever and tone arm combination, is improved. I believe this would help to also recover transients in the recording since the stylus is being dragged slower across the groove surface. What I really can't deal with even with a heavy platter direct drive turntable is the shift up of the turntables rumble frequency specifications which generally are below audiblity but at twice the frequency, it's too audible for my tastes and I don't want to heavily filter out those lower frequencies, since it is then in the audible bandpass. So it's all a big push and pull compromise. So, I'll just deal with the 78 rpm playback speed that can be then tweaked in real time for proper key signature/pitch of playback. We all know that 78 rpm recordings were never 78 rpm but averaged between 72 to 84 rpm on average.

Then onto the noise reduction of these old beat up, overplayed 78's. Recording them to a stereo track reveals that the clicks and pops are different, in both the left and right channel. So even though these are mono recordings, I'll transfer them as stereo, reduced the clicks and pops in stereo, then collapsed to mono. The noise reduction seems to do a better job of that than trying to accomplish the same thing in the file is merged to mono first. Then surface noise. Then frequency response. I'm using Adobe Audition's noise reduction features, Sony/Sonic Foundry Noise Reduction and MAGIX cleaning lab. Not altogether necessarily a dependent upon each disks condition. A lot of time will be spent on trying to recover the best of these old historic recordings of this late 1940s Metropolitan Opera star.

The first of my New Year's resolutions is to start this project.
Ms. Remy Ann David

x

User login