Print

Print


Hi, Scott,

As you know, when using DAW speed correction, there are several 
approaches. There are algorithms that will adjust the speed without 
altering the pitch. There are algorithms that will change the pitch 
without altering the speed. Then there is resampling.

Do you recall which type of algorithm you used that caused you to note 
that they sound worse when stretching?

I have never found any of the single dimension algorithms to be very 
useful for major speed corrections. I only use resampling and then it is 
essentially as smooth as the mechanical means using a speed adjustment. 
Of course, if one is slowing down the recording, you don't want to run 
out of samples, so even if the bandwidth of the reproducer doesn't go to 
100 kHz, if the speed change is 2x (12 semitones, 1200 cents) then doing 
the original transfer at 96 kHz still leaves 48 kHz effective sampling, 
but if it's 4x (24 semitones), you'll need to transfer at 192 kHz.

It may be better to transfer at a slower speed and speed things up to 
not limit the bandwidth. EQ is then a bit of an issue, but guidance can 
be found in MRLtapes.com 's "Choosing and Using..." document where they 
show corrections for using a calibration made for one speed at a 
different speed.

My two primary software tools both can do this effectively:
Samplitude (also Sequoia) provides "Elastic Audio" and will behave in 
this manner if the "Resample" algorithm is selected.
Izotope RX7 provides "Variable Pitch" and will behave in this manner
if the "Preserve Time" checkbox is UNCHECKED.

Synching this pitch change beyond two tracks (stereo) in either 
application will require some experimentation. The one time I had to fix 
a four-channel tape that had 4x speed correction, I copied it to a 30 
in/s 4-track 1/2-inch tape and varispeeded it down to 7.5 in/s as the 
tape was playing. I had to stop in the middle and change playback speeds 
to continue.

Again, one person's experience -- there are many ways to do this.

Cheers,

Richard





On 2020-03-05 8:30 p.m., Scott D. Smith wrote:
> Parker:
> 
> This would depend in part as to whether the varispeed is constant across 
> the length of the tape, or if it varies. It would also depend on how 
> much it varies.
> 
> In general, I prefer to maintain a constant speed for playback, but if 
> the subsequent files need substantial alterations to get them back to 
> "normal" playing speed, then the digital artifacts can become extremely 
> annoying (especially if stretching instead of compressing the length)
> 
> I have had to deal with some tapes that were recorded on battery 
> operated equipment where the speed has varied by more than 20%, due to 
> the batteries going dead during recording. This required some rather 
> severe varispeed correction (beyond the range of what most standard 
> professional machines can do). In general, I have favored doing gross 
> speed corrections in the analog domain, and final tweaking in the 
> digital realm.
> 
> In some cases, we have made one pass at constant speed, and another with 
> varispeed correction.
> 
> In any event, any corrections should be noted as part of the subsequent 
> file metadata.
> 
> This is just my individual approach to the issue. To the best of my 
> knowledge, there is no "standard" best practice for this.
> 
> Scott D. Smith CAS
> Chicago Audio Works, Inc.
> 
> 
> 
> On 3/2/2020 4:58 PM, Parker Fishel wrote:
>> Hi ARSC listers,
>>
>> We have some 2” session tapes that we are currently getting digitized. We
>> encountered some varispeed reels and are trying to figure out how to
>> properly transfer them for preservation purposes. What are the best
>> practices? I did a little digging around, but haven’t been able to 
>> come up
>> with much.
>>
>> Thanks in advance.
>>
>> Kind regards,
>> Parker Fishel

-- 
Richard L. Hess                   email: [log in to unmask]
Aurora, Ontario, Canada                             647 479 2800
http://www.richardhess.com/tape/contact.htm
Track Format - Speed - Equalization - Azimuth - Noise Reduction
Quality tape transfers -- even from hard-to-play tapes.