recording from analogue cassettes

Discussion forum for Amadeus users

Moderator: Martin Hairer

Post Reply
Tom D
Posts: 4
Joined: Wed Jun 15, 2011 8:55 am

recording from analogue cassettes

Post by Tom D »

I am using an ION Tape 2 PC to play numerous C90 tapes of spoken voice to digitize the data.
I plug into the USB on my Mac and use the 'Audio CODEC' facility which seems to give the right level of sound input (I find that if I set input manually and the voice changes suddenly - as it does - I get much clipping so decided it is best to use this audio CODEC facility).
These voices were made using a cheap hand held portable device that took standard cassettes, which is what I am playing in real time to make the transcription.
They are therefore of poor quality. However, they are for archive so I am recording them to WAVE format (before making an mp3 for transcribers to use).
Obviously these are mono tracks.
When I make the recording I use the 'new file' facility in Amadeus and play the tape into the computer, which I note gives me a representation of the track as if stereo on my monitor.
The bit I am puzzling over is that when I check the sound character for these two 'channels' (top and bottom) I find they are *different* and one looks bigger than the other (the top being visually fatter than the bottom).
I have tried to find if there is a facility to set both 'input channels' (if you like) so they are exactly the same but the USB CODEC function does not seem to have a setting like this.
Why is this apparent discrepancy occurring and does it matter?
Thank you.

Sonic Purity
Posts: 82
Joined: Sat Nov 10, 2007 11:58 pm
Location: Pasadena, California, U.S.A.

Re: recording from analogue cassettes

Post by Sonic Purity »

Tom D wrote:I am using an ION Tape 2 PC[…]
I plug into the USB on my Mac and use the 'Audio CODEC' facility[…]
These voices were made using a cheap hand held portable device that took standard cassettes, which is what I am playing in real time to make the transcription.
[…]
Obviously these are mono tracks.
Actually, that's not obvious, so it is good that you mentioned it. Cheap handheld stereophonic cassette recorders were made at one time.
Tom D wrote:When I make the recording I use the 'new file' facility in Amadeus and play the tape into the computer, which I note gives me a representation of the track as if stereo on my monitor.
It is. The Ion is a stereo cassette deck. Neither the Ion nor Amadeus can tell whether the actual cassette is stereophonic or monophonic. (Nor can any other software or hardware i’ve heard of, for that matter.)
Tom D wrote:The bit I am puzzling over is that when I check the sound character for these two 'channels' (top and bottom) I find they are *different* and one looks bigger than the other (the top being visually fatter than the bottom).
I have tried to find if there is a facility to set both 'input channels' (if you like) so they are exactly the same but the USB CODEC function does not seem to have a setting like this.
Why is this apparent discrepancy occurring and does it matter?
Thank you.
I’ll answer backwards:
Does it matter?: No.

Why?: The actual magnetic recording track on the physical tape is fully covering the left channel playback head in the Ion, but not fully covering the right channel. Or, it is fully covering both but there is more fringe effect on the left channel. In either case more signal on the left than the right.

This is not at all unusual when a true, physical, single-track monophonic tape head made a recording and it is being played back on a physical double track stereophonic tape head.

I suggest not bothering getting the levels to match. You might even consider having the two channels a safety factor, or redundancy: if there is a momentary drop-out, it may only affect one of the two channels, not both. In that case, you will be able to copy over the solid signal from the one channel to the other.

I strongly suggest that, unless you don’t have time and are pushing off handling the archives for the future, that you carefully listen and select one of the two channels to keep, whichever one is best (best quality, fewest drop-outs). If you want to fill in drop-outs from the other channel, use the Amadeus Pro “Amplify” function to raise the level of the right channel to match the left channel (having first selected the whole of the sound file and looked at Waveform Statistics [i’m doing this from memory… that was the Amadeus II name… it may have changed in Pro] and having done the math to find out the numerical level difference) so that they’re very close to the same level, then copy over the dropped-out bits as needed.

You should now have one channel of the digital file which is the best quality and has its drop-outs filled in. The other channel is now superfluous. Select all of this other channel, delete it, then convert the sound to mono (or take other actions which you may prefer to wind up with just the one channel). You now have a master file, truly mono, as good as it will get in quality (short of messing with digital enhancement tools, which can be done with copies of this master file later on). It will be approximately half the file size of the stereo (well… two channels of not-quite-exact mono sound… stereo in file structure and the space it takes up) file originally recorded.

Note that you should not sum the two channels together with tape formats, especially cassette! Unless by chance the azimuth of the playback tape head exactly matches that of the recording play head (and chances of this are very, very slim), you’ll get unwanted high-frequency cancellation effects which in most cases will seriously degrade the sound further. Summing is usually appropriate for mono phonograph records played on stereo hardware and mono content FM broadcasts received as stereo, but not for tapes. You might actually want to make a copy of your archive master and try going against this advice (with the copy) to hear what happens... it may be instructive.
))Sonic((

Tom D
Posts: 4
Joined: Wed Jun 15, 2011 8:55 am

recording from analogue cassettes

Post by Tom D »

Thank you very much to Sonic Purity for this very helpful and highly informative post. Much appreciated.

Tom D.

Tom D
Posts: 4
Joined: Wed Jun 15, 2011 8:55 am

Soud levels for distant voices

Post by Tom D »

I think I have not articulated my concern correctly so have edited my question and a bit of text below.

I realize this post has overlap with 'Cassettes to MP3 transfers for Dummies' thread. However, my question is specific.

As described above I am transcribing elderly cassettes of audio voice to WAVE format for eventual archive.

At the moment I am not particularly interested in cleaning these recording up or adding markers etc. I am simply playing the tapes in real time and making 'master wave' files for future eventual archive.

However, I (obviously) need to be sure I am doing this transfer right.

The latest issue is that some of the tapes have 'distant voices.' In other words the tape recorder has been placed on a table and folks talk over it. The voices are therefore 'distant' and there is a lot of background hiss.

To date I have been using the Audio CODEC facility to set 'correct' sound levels for input because I have found this works well for tapes that are basically dictations where the voice is close to the mic and very clear (if rather poor quality with highly variable sound levels) . [as mentioned above I find setting the manual input can result in a lot of clipping as sounds vary wildly, which is why I am using the Audio CODEC facility].

However, with these 'distant voices' tapes I find I end up with a 'trace' that is quite 'flat' but if I turn to manual input and crank up the level, the recording seems to sound worse even though the trace is larger because when I amplify the voice I also seem to amplify the background hiss (I think Martin mentions this effect somewhere in his manual).

Obviously I can use the 'amplify' function after the transfer to make the sound louder. I could also, instead, increase the input (while carrying on using the CODEC facility) by simply turning the input up manually on the back of the Ion.

Question: is there a way to know what level of input to use to maximizes clarity of 'distant voice' recordings relative to background noise (hiss) or do I just make the transfer into digital, not worry too much about getting a flat trace (these voices are distant after all) and then deal with the hiss and initial sound as it is afterwards (using de-hiss, amplify, etc.)?

[Under this option I assume I set the input manually from the ION based simply on how the sound is to my ear (as stated if I ramp up the input I find the background also ramps up too)].

I hope that's clearer.

Thank you. Tom D

Post Reply