Digitizing cassettes - question regarding levels

  • Thread starter Thread starter alaurence
  • Start date Start date
A

alaurence

New member
Hi,

I'm digitizing just over 100 cassettes for someone for archival purposes. I have an okay setup at home and they didn't want to pay for a premium service, so I'm charging a bit less to do it.

I've got some questions about levels.

First, my setup: Kenwood KX-550HX > Focusrite Scarlett 2i2 (using a dual RCA to 1/4 inch cable plugging into the input) > Ableton Live 10.
I also have the Waves WLM Plus metering plugin, and intend on using the Waves Restoration Bundle in post.

1. When deciding on a good level to set the gain on the 2i2, I understand this is a bit of trial and error, making sure the level is flashing green but not going into overdrive by hitting yellow and red. One thing I wish the 2i2 had was numbers around the gain knobs, so I could remember where I've set it for this work. Any suggestions other than drawing on it?

2. These are the levels I'm getting (and for reference I've got the gain knobs on the 2i2 set at about 11 o'clock):
  • Hovering between -14dB and -6dB when looking at Ableton's built-in metering.
  • When looking at the WLM plugin, I'm getting
  1. Short term: between -18 and -21 LUFS
  2. Long term: -20 LUFS
  3. Range: 7 LU
  4. Momentary: -10.1 LUFS
  5. True peak: -1.5dB

Thoughts on this? Obviously my aim here is to record this in to Ableton with as little tampering as possible, so I want to make sure it's not too high or too low (on the 2i2).​

3. Any other general advice when digitizing cassettes?

Thanks!

EDIT: Last question - I'm currently using the front inputs on the 2i2, but my dad has a 4i4 I could borrow. Would it actually be better to use the fixed-gain inputs on the back of the 4i4 instead of having to adjust the knobs on the 2i2?
 
Last edited:
First of all, I would be recording at 24 bits. That gives you more than enough headroom that you could record at -30dB peak, then normalize it to around -1dB in post and never lose a bit of quality. Cassettes don't have a tremendous dynamic range, so you're putting a 1 gallon can in a 5 gallon bucket. Everything is going to fit, even at 16 bit. There's no way you will increase the noise floor with the transfer.

The numbers you show are perfectly fine. You don't want to exceed true peak, so if you want the overall sound to be "louder", then you're going to have to limit the peaks (there may just be a few-- check the waveform) and raise the level. You can either use a limiter or a compressor.
 
If you're getting peaks above -30 dBFS and below 0 dBFS, you're fine. Honestly, if you're recording in 24 bit, you could have peaks at -45 dBFS and still have the cassette noise floor above the digital noise floor. I use a first generation Scarlett 6i6 to digitize stereo and 4-track cassettes. I find that the front inputs set to the noon position match the rear fixed-gain line input sensitivity. A different generation of Scarlett may differ in that respect.
 
Also you just need to watch the actual record level of home recorded cassettes it can vary immensly. You need to bring any extremes back into control.
 
bouldersoundguy said:
If you're getting peaks above -30 dBFS and below 0 dBFS, you're fine. Honestly, if you're recording in 24 bit, you could have peaks at -45 dBFS and still have the cassette noise floor above the digital noise floor. I use a first generation Scarlett 6i6 to digitize stereo and 4-track cassettes. I find that the front inputs set to the noon position match the rear fixed-gain line input sensitivity. A different generation of Scarlett may differ in that respect.
Click to expand...
Thanks for this. Now I'm wondering: my dad has a 4i4 second gen I could borrow - would using fixed gain inputs on the back of that be preferable to what I'm doing right now using the front inputs?
 
rob aylestone said:
Also you just need to watch the actual record level of home recorded cassettes it can vary immensly. You need to bring any extremes back into control.
Click to expand...
A friend suggested using the Waves Vocal Rider plugin which adjusts gain in real time. Worth using? And at what settings if so?
 
alaurence said:
Thanks for this. Now I'm wondering: my dad has a 4i4 second gen I could borrow - would using fixed gain inputs on the back of that be preferable to what I'm doing right now using the front inputs?
Click to expand...
It would be slightly simpler. Technically, it would have less circuitry in the recording path. I found that the rear inputs were just right for consumer line level, and the level never got near clipping. But you can get pretty much the same result with the front inputs.
 
alaurence said:
A friend suggested using the Waves Vocal Rider plugin which adjusts gain in real time. Worth using? And at what settings if so?
Click to expand...
No, this is a bad idea to use while recording, but after it's recorded you could try it to see if it does something useful for you. That plugin will turn up parts of a song that are supposed to be quiet and turn down the parts that are supposed to be loud. It can work on spoken word, but it can do strange things depending on how it was recorded.

You simply need to set the input levels so it doesn't clip the input of the interface. Since you don't have to get anywhere close to clipping to get a good recording, just set it so the peak levels are no higher than -6db. Capture the audio as is, then you can play with it to make it 'better'.
 
When we were recording LPs onto cassettes, thye system we used was a bit hit and miss. You'd dib in and out of the LP looking for loud parts and then you'd decide how far to push the meters. CrO2 tapes were a favourite of mine and you could really push the levels without much consequence and get better noise performance. On playback, you'd often have to turn the volume down - so this was quite different from proper lineup where you would have a tone to follow. All you need is the loudest bit and pick your own maximum digital level and adjust the loudest part of the recording to match that and leave the dynamic range as wide as it can be, Squashing it even more is a bad move.
 
rob aylestone said:
When we were recording LPs onto cassettes, thye system we used was a bit hit and miss. You'd dib in and out of the LP looking for loud parts and then you'd decide how far to push the meters. CrO2 tapes were a favourite of mine and you could really push the levels without much consequence and get better noise performance. On playback, you'd often have to turn the volume down - so this was quite different from proper lineup where you would have a tone to follow. All you need is the loudest bit and pick your own maximum digital level and adjust the loudest part of the recording to match that and leave the dynamic range as wide as it can be, Squashing it even more is a bad move.
Click to expand...
Cool, so practically speaking you mean just make sure the loudest part/s aren't clipping and then hit record?
 
Basically that, yes. The snag with cassettes is always noise and the general audio quality, so if you have 100 to do, you can make sure they all can be played back without needed to adjust the volume knob, which I bet you have to do a lot unless they are pre-recorded ones that followed a proper rule for levels.
 
The most important thing to do when digitising cassettes is to match the azimuth angle of the playback head to the tape. Nearly every tape is different and, if you don't do it, you may lose information that cannot be recovered later. Some expensive decks have a front panel azimuth control but every cassette recorder that I have ever seen has an azimuth screw that can be accessed easily. You may have to remove the trim from the cassette door or, if you are lucky, you may find a small screw hole in the cassette door. You can find a basic guide to adjusting the azimuth at

thegreatbear.co.uk

azimuth adjustment when you transfer and convert cassettes to cd | greatbear audio and video digitising

Cassette tapes run at a very slow speed of 17⁄8 inches per second (ips) with a very small track width of 1.59mm Cassette decks when they left the factory or a service centre should have been aligned…
thegreatbear.co.uk thegreatbear.co.uk

although there are variations on this depending on what gear you have available.

You can sometimes fine tune the azimuth (or, more accurately, the delay between channels) in software but you can't compensate for more than a very slight adjustment so it is vital to get it right before transferring the cassette to digital.
 
Almost word for word my standard advice on head alignment.^^
 
What others have said. I just digitized a cassette that was never released on CD. Set the levels so peak are below say 3dBFS and record at 24 bit. I'd also set sample to 44.1kHz if you think you (or someone) will be burning CDs, since it will save a resample step later.

I would look into RX for noise reduction. Just a little sample of the "silent" part of the tape can be used in "LEARN" mode of its "spectral" noise removal, and I found a light touch with that is really nice to clean things up. Maybe it's not an issue for this job, but I liked the results.

Opinions differ, but after I got the capture, I "normalized" the results. You have to try and decide if differences in loudness on the cassette "tracks" are something that matters, i.e., it was carried over fairly intact from the original LP, or if it's some artifact of any "mastering" done for the cassette or its manufacture; or, maybe differences in how well your tape deck aligns to sides 1 and 2 (I would not use auto-reverse, but that's me). In any case, you can either normalize (with gain only!) everything to level out the peaks, somewhere in that -1.0 to maybe -0.5dBFS, or normalize loudness (what I did), and let the peaks float, using the loudest track to determine where you can push things to. (All IMHO, of course.) Me, I'd want to keep overall loudness appropriate for the genre, regardless of peaks, since the contents may have already been pretty compressed.

I also dithered (another opinion) and exported to 44.1kHz/16-bit. I can't imagine that there is dynamic range on a cassette that needs 24-bit, even after Dolby and/or noise/hiss cleanup.

p.s. clean those heads often!
 
Cassettes can be amazing quality but often terrible too. Ferric oxide was so random. There were tapes that were stupidly cheap and others that were much better. The standard of the recorders changed as did the mechanical aspects. It means that a box of cassettes will always contain surprises.
 
You must log in or register to reply here.
Back
Top