How to clean and master an MP3 with the following sound distortions?

karimsisi wrote on 7/1/2015, 7:40 AM

Hello,

I need help with an issue I can’t solve. I am a beginner at digitizing audio. However, I successfully managed to digitize a cassette tape recording I had (including cleaning, mastering, and exporting) using Magix Audio Cleaning Lab within 3 days of my first use due to its ease of use; Audio Cleaning Lab is a work of art!

The issue I am facing is with a half-hour speech that was recorded back in the 80s. I listen to this speech on my car’s mp3 player, but due to the poor quality of the recording, I have to make the following modifications to the car’s mp3 player settings to hear the speech clearly:

Lower the bass to the minimum because the speech is muffled.
Increase the treble all the way to make the voice thinner.
Increase the volume to the maximum to hear clearly because the volume of the recording is too low. The problem is when the speaker raises his voice every few minutes to emphasize a point, his volume becomes very high for a few seconds and it hurts my ears before it goes down again to the regular low volume.
There is an artificial sound to the speech that I can’t really describe, as if there is a short echo, or the recording was made in mono and then artificially converted to stereo and one channel is not in sync with the other by part of a second (I hope I am using th correct terminology here).
The Ds and Ts of the speaker sound the same due to the poor quality of the recording, and whenever, he pronounces a word that requires air exhale, such as “aha” the recording is flooded with a noise that overwhelms any other noise and I find it difficult to understand that word.

When I imported that speech, which was already converted to an mp3 file, into Magix Audio Cleaning Lab to attempt to clean and master it, I failed. I was successful, however, at recording, cleaning, mastering, and exporting a 20 year old cassette tape recording. I was successful because I went through each and every cleaning and mastering setting until I achieved the required voice quality. Alas, I failed to clean and master the speech when I went through each of the cleaning and mastering settings like the 20 year old cassette tape recording.

I noticed one more thing when I zoomed into the speech’s wave pattern, it wasn’t smooth; it has jagged edges like a bar graph. This was not the case when I compared the cassette recording I digitized; the zoomed in wave pattern was smooth without jagged edges. Could this mean the mp3 file (speech) was already worked on by a digitizer? (I downloaded the speech as an mp3 file from the internet).

Attached below, are screenshots of the speech and cassette recording zoomed wave patterns.

Screenshot of speech with jagged wave pattern

Screeshot of cassette recording with smooth wave pattern

Here is a 20 second portion of the audio file

If you could explain in simple terms what are the problems this speech suffers from, and what settings to use to clarify this speech, I would be grateful. One final note, this speech is a lesson and I have hundreds of lessons with the same issue, is there a way to setup a template with the necessary cleaning and mastering settings and then batch process them?

Looking forward to hearing from you soon.

NB. I am using Audio Cleaning Lab 2016 (not premium). Would using the premium version help better?

Comments

johnebaker wrote on 7/1/2015, 9:50 AM

Hi
The problems you are experiencing are very common when trying to capture and clean old tape recordings.

Lower the bass to the minimum because the speech is muffled.
Increase the treble all the way to make the voice thinner.
Increase the volume to the maximum to hear clearly because the volume of the recording is too low. The problem is when the speaker raises his voice every few minutes to emphasize a point, his volume becomes very high for a few seconds and it hurts my ears before it goes down again to the regular low volume.
There is an artificial sound to the speech that I can’t really describe, as if there is a short echo, or the recording was made in mono and then artificially converted to stereo and one channel is not in sync with the other by part of a second (I hope I am using th correct terminology here).
The Ds and Ts of the speaker sound the same due to the poor quality of the recording, and whenever, he pronounces a word that requires air exhale, such as “aha” the recording is flooded with a noise that overwhelms any other noise and I find it difficult to understand that word.

1,2 3 are due to the loss in strength of the magnetic fields in the tapes particle coating, and the loss varies along the tape.

4 Is caused by what is called 'Print through' the magnetic field from 'one layer ' of the tape, as spooled, affects the adjacent layers and produces faint 'copies' of the information it holds.

5 Is due to the loss of magnetic strength and the increased tape hiss level of the tape due to age, this will be worse in sections of the tape where the magnetic strength of the recording is weakest.

. . . . when I zoomed into the speech’s wave pattern, it wasn’t smooth; it has jagged edges like a bar graph. This was not the case when I compared the cassette recording I digitized; the zoomed in wave pattern was smooth without jagged edges. . . . .

This is overload distortion - whoever did the recordings has either not been careful to set the recording parameters correctly ie they are set the recording level too high, or has over processed the audio - I would suspect Normalisation has been applied with the incorrect sound level parameters.

The bad news is once this happens there is very little you can do - it is not possible to put back what is missing, many audio cleaning effects involve the removal of audio information of one kind or another - eg tape hiss. Using effects such as boost, eg bass, treble etc, also increase the unwanted noise and losses.

If you have tapes that are 'younger' than the above it would be wise to digitise them before they too suffer any further losses.

John EB

karimsisi wrote on 7/3/2015, 9:43 AM

Thanks johnebaker for your comprehensive reply :)

Before posting my question, I didn't know I could attach audio files, so I edited my question and included a 20 second clip of the file. Would it help if you heard it?

Thanks again.

browj2 wrote on 7/3/2015, 11:42 PM

Hi,

Does the 20s clip sound like that when you download it from the internet? Or is it after you ran it through ACL?

Can you give us a link to a file in Dropbox or some other public access system where we can download an original version?

If it comes like that from the source, then it is badly messed up. I can make good sound files sound bad like that, but going the other way is tough.

karimsisi wrote on 7/4/2015, 1:45 PM

@browj2.

The 20 second clip is without any editing on my side. It sounded like this when I downloaded the 30 minute speech (MP3) from the internet.

Can you give us a link to a file in Dropbox or some other public access system where we can download an original version?

Here is the link to the MP3 file I downloaded https://drive.google.com/file/d/0BzJjBsmVLNxRNVZ6cFJlLVVJYk0/view?usp=sharing

johnebaker wrote on 7/4/2015, 5:21 PM

The audio file is not showing any overload distortion, however I suspect it has been :-

- heavily denoised, probably using multiband filters

- a noise gate has also probably been used

There is no way of cleaning the file up - the damage has already been done.

John EB.