This is the next instalment of our audio home quiz challenge exploring quality control of digitised audio.
WARNING: this is much tougher!
How well can you identify these 5 issues within digitised audio?
(NB: won’t work in Internet Explorer sorry!)
Question 1:
EXPLANATION:
After you’ve answered listen here:
As the power in the batteries drains, the capstan motor turning the tape in the recorder is slowing down. This means that increasingly less tape is being passed over the record head as the audio signal is being recorded. Conversely when you play this tape back on a tape deck with full power, the capstan motor is running as normal and the tape is being played back at the correct speed. So this means that the original signal goes over the playback head and sounds much faster to our ears as it was recorded on a shorter portion of the tape, which creates the effect of it speeding up rather than slowing down.
Question 2:
EXPLANATION:
One of the most important parts of the digitisation of analogue recordings is to clean and maintain your playback equipment. Tapes which are sticky leave lots of gunk that builds up over the rollers, tape lifts, and heads. The playback head needs to be spotless to have the best opportunity to capture the full signal. Archival best practice means we’re trying to do everything we can to get the best signal on playback, so we clean our machines between every reel.
In the worst cases we even bake the tape overnight at 50°C in a conservation (not domestic!) oven. This drys out the moisture to give an opportunity for improved playback the next day. The British Library have many years of experience baking tapes and for this project provided all the UOSH hubs with a tried and tested food dehydrator. 12 hours on the mushroom setting works wonders!
Question 3:
EXPLANATION
This was a tough example! This dropout is an analogue drop out on 1/4 inch tape. Looking at the Spectrogram above you can see a gap in the audio. There are straight vertical lines to show this but importantly these lines don’t go all the way the top of display, meaning that they don’t go through all the frequencies. This is the first clue that this is not a digitally introduced dropout. Playing the tape back to listen again would be essential here.
Below is a spectrogram showing a dropout created unintentionally during the digitisation process . The line is abrupt and vertical, covering all frequencies. The second image is zoomed in on this same scope:
This type of error is introduced and not in the orignal source. It can also sound like a gap in the audio so it’s important to QC your digitised files to check:
Analogue audio has a smooth waveform curve. When we digitise audio at archival standards (96kHz 24bit) we are trying to best represent that analogue curve. Sometimes there can be a glitch that can be introduced during digitisation, usually due to the performance of the computer. This results in dropout (or interstitial error) on the exact timecode of that glitch, whereby audio is lost in the file. This means that the digitised file is not a good representation of the original analogue recording and requires digitisation again.
Conversely if the loss of audio is in the original analogue source, then the resulting digital file is already an accurate copy and would not need to be redigitised. To confirm this, you should play the source again and double check by listening and looking at the same portion of the tape. Compare to the file you had already made. Does it sound the same? If the heads on your machine are clean then you can confidently assume this error was insource and therefore should be retained with in the preservation file.
Question 4
EXPLANATION
Audio A is the result of an open reel tape being played on the wrong side of the tape. This is the backing of the tape and not the surface where the audio signal is held.
Audio B is the same audio laced up correctly with the oxide side facing the playback head. This is the side that carries the audio as magnetised particles.
Question 5 (the final question!)
The following audio file was created by Indiana University:
As Indiana University explain:
‘The track begins before the azimuth is adjusted and you will hear the following:
About (0:06)—the adjustment is made. You will hear increased clarity and an increase in level (volume).
Around (0:15)—azimuth is set back to the original (incorrect) setting
Around (0:23)—azimuth is adjusted again.’
EXPLANATION
Changing the azimuth is like the audio equivalent of focusing a lens. Instead of either side of that focused point being blurry, with audio it sounds muddy. This is because the higher frequencies are not being picked up and made audible.
The angle of the recording head can be wildly different on recording devices. So when it comes to playing these tapes back you can make an adjustment on the machine to reposition the head to match the same angle on the recording. That is what’s happening 6 seconds in. Someone is fine-tuning the azimuth angle so that all of the signal available can be heard.
After cleaning and maintaining playback equipment, setting the correct azimuth for each recording is one of the most important things to do when digitising audio. We play the tape, listen carefully and use audio scopes to make sure the azimuth is correct. If all the other settings are also correct we then rewind the tape and then we’re ready to record with the correct azimuth set from the start.
Above is an example of what a Phasescope looks like with 4 recordings running all at once.
Above is an image of an azimuth modification dial added to a Studer Open-reel machine.
Above is an image of a hex screwdriver being using to adjust the azimuth on a audio cassette tape deck with a modified hole for access.
YOU’VE FINISHED THE QUIZ!
For further info please check out these excellent resources:
IASA Special and Technical Publications
For more on the British Library’s Unlocking Our Sound Heritage project: