Family History Cassette Tape Project - Tips to Begin?

Deborah-Wesley wrote on 2/7/2025, 6:47 AM

I am new to Audio Cleaning Lab 4 and know very little about audio editing. But everyone has to start someplace. I have about 200 cassette tapes from the 1980s and '90s of family interviews and conversations that I am going to digitize. I have established my process to record to wav files - that is going well. (I am recording in mono.) My end game will be to use some of the audio in family history projects (such as setting the voice to photos.) I also plan to transcribe some of the tapes using a service such as Descript. My set up is a Windows 10 desktop - rather old (13 years) but has an i7 processor and 32GB of memory; I would love a new set up - but it's not going to happen in time for this project.

Where I could use some advice is on the cleaning portion of the process. Frankly, I am a bit overwhelmed by all the choices. The largest issues with the recordings are:

Rumble, hum, and/or hiss from the tape recorder
Ambient noise (such as car noise while driving to a cemetery or wind)
Uneven volume as people move about the room, or voices drop off

I am starting to play with the various options under Cleaning and Mastering, I see besides the basic options there are a plethora of plugins as well. Is there someone who could point me to the ones I should concentrate my learning curve on.

With 200 tapes to process I don't need perfection. Afterall, I have to get back to photo scanning (standing at 18,000 photos scanned - perhaps a third of the family collection.)

Comments

SP. wrote on 2/7/2025, 7:46 AM

@Deborah-Wesley To reduce uniform(!) noise like tape hiss, I recommend you use the 32 bit version of Sound Forge Audio Cleaning Lab 4 because the normal 64 bit version always throws annoying error messages. Technically it works fine, but it's always extra clicks.

To isolate voices from other noises automatically, you might try a free modern AI based solution like Ultimate Vocal Remover which can split voices from background noises and music (that's its original use case). If your computer has a Nvidia graphics card (even an old one like a 1660), this will also work relatively fast. It also works with an Intel CPU but will take a lot of time, probably 1:1 (1 minute of audio needs 1 minute of processing).

This should give you somewhat clean audio for people speaking. Maybe test this with some short audio snippets.

Besides removing uniform noises and isolating voices it gets tricky. You usually need to do manual editing, for example in the spectral editor. But this can become a monumental task. It would look somewhat like this (be aware this is an old video for Magix Samplitude, but spectral editing is also available in Sound Forge Audio Cleaning Lab):

browj2 wrote on 2/7/2025, 9:53 AM

@Deborah-Wesley

Hi,

I'm doing pretty much the same thing but I also have 8mm, Super8 with sound film, VHS, and other video files to go along with digitized photos and cassettes. You are undertaking an enormous project so I suggest that you get the entire work flow figured out before going much further. File management and documentation (register/database of projects, digitized and exported files) is one of the very important elements as you will see as we go through this together. I'm still sorting this out and suffering from not having thought through the entire process.

I hope that you have been digitizing to wave files and not mp3.

Before going further, @SP. mentioned the error (101) that SFACL4 throws up at the beginning and when doing certain things. It seems that no everyone has this problem. Do you have this problem?

Next, SFACL4 comes with a lot of additional plugins. Did you install and activate all of these?

Most importantly, you should have both the iZotope Elements 10 and SpectraLayers Elements 10 installed and activated.

There are many cleaning tools out there that you can use. iZotope and SpectraLayers have similar tools, but SpectraLayers has some features that are very interesting, like unmixing vocal (voice) from the background, allowing you to clean/reduce/modify/eliminate background and other noises.

As @SP. showed, you have the Spectral Cleaning native to SFACL4 and it's very good for removing specific noises and glitches. You can do the same in SpectraLayers, but it's more complicated.

There is also Accusonus 6 cleaning tools that are very good. There are a couple more tools than the others propose, including DeBreath and PlosiveRemover. These were made available for free but eventually removed. If you want these, send me a personal message.

Next, you may want some better tools for mastering, in particular compressor and equalizer. I suggest that you download and install the free bundle from Melda.

Once you have everything set up, we can look at what to use, how and when.

Out of curiosity, which video program are you going to use?

John CB

Deborah-Wesley wrote on 2/7/2025, 12:41 PM

@SP. and @browj2 - Many thanks for your replies. I do indeed get the 101 error. I tried using the 32-bit version instead, but then when running Auto Cleaning and some other functions I received new error messages that I suspect were memory related. So I have reverted to the 64-bit version. Which does seem to work. I replaced my graphic card last summer with a GeoForce RTX 3060, hoping to buy a couple years until the inevitable PC replacement.

@SP.- Thank you for the suggestion of Ultimate Vocal Remover - I will give it a look. As to manual detail editing - I will only revert to doing so if I have a VERY high value tape with MAJOR issues. Considering the wealth of the family archive I have amassed, I have to adhere to the 80-20 rule (or maybe 90-10).

@browj2 It is nice to "meet" someone who is traveling down a similar path. I have three closets and four file cabinets stuffed with the family archive - photos, audio, ephemera, and - you are correct - video. (Bless my supportive husband.) Several years ago, before I retired, I had the old 8mm film digitized by a service - the result is mediocre. I have transferred many of the VHS tapes to DVD. The whole audio project needs a lot of help, and rework, but that is a challenge for another day (or more accurately, another year.)

I certainly echo your sentiments that a documented workflow with an organized file system is essential. Perhaps more important is making sure a solid backup plan is in place. (I was an IT professional for 40 years, hence I have four copies of all my data - my primary NAS, an onsite backup drive, an offsite backup drive, and a cloud backup.) Plus subsets of the data shared with cousins occasionally.

I appreciate any and all tips you can share on this process. I am glad to learn from the hard-knocks of others. I am saving to .wav. Again, I know virtually nothing about audio (yet), but I relate wav / mp3 to the tiff / jpg image standards. I do have some mp3 files that I have gathered from other folks, but the cassettes I have started recording are all to .wav.

101 Error: As I mentioned, I do have the 101 error. Are you running 32-bit or 64-bit? If 32-bit is working for you perhaps I should give it another try, but I seemed to have other issues with the 32-bit.

Plugins: I did not have the plugins until early this morning. I had seen links under Cleaning to the plug-in modules, but then it asked for a license key. It was not until last night as I was reading some of the forum messages here, that I realized I had the license keys - and more download links - in my email. So I got those all installed and registered this morning. These extra plugins are part of what prompted me to post here - I think I could spend the next month, if not more, learning about all the cleaning and mastering options. While that would be fun in its own right, I need to get on with this project so I can get to the end game - using these digital assets to create material that will share and preserve our family history with the next generation. I have a LONG list of ideas, longer than I will ever get done. So I don't want to spend unending time on just figuring out the process. (Although, as we both have said, getting a good process in place is critical.) So I installed / registered - wizardFX MAGIX Extension Pack, iZotope RX 10 Elements, iZotope Ozone 11 Elements, and SpectraLayers Elements 10 - although SpectraLayers may be version 11.

So with all these plugins, where would you suggest I start cleaning? I am sure the answer is "it depends", but where would be your first go to tool for: the following three items that run throughout the tape? I think the first two are part of Cleaning, and the third part of mastering. Is that correct?

Rumble, hum, and/or hiss from the tape recorder
Ambient noise (such as car noise while driving to a cemetery or wind)
Uneven volume as people move about the room, or voices drop off

I will look into Accusonus 6 and Melda if you feel those would be helpful beyond what came with Sound Forge Audio Lab, especially since the price is right.

Video program is a good question. For many years I have used ProShow Gold by Photodex. Photodex ceased support in 2020, but the programs still run on my PC. I doubt though that I will be able to install and activate when I do replace my desktop. A couple years ago I purchase Davinci Resolve when they were practically giving it away. I have played with a bit but have not created a major project with it yet. What do you use?

For photos I live in Lightroom, augmented by Photoshop and Topaz AI.

If you have made it to the end of this (I did not intend to write so much), thank you.

Debbie

browj2 wrote on 2/7/2025, 5:28 PM

@Deborah-Wesley

Hi Debbie,

Good stuff.

To get you going, open a project or record.

The first step is to get a good recording with the volume set so that you don't get clipping - nothing that goes up to 0 dB, with most between -20 to -10 dB.

Take a look at the waveform. If it is large with no spikes going to the top, then your recording volume setting was probably good. If it's small all along, then you recorded too low and may want to increase the record volume in the Windows settings.

Next, I Normalize. This is a contentious issue as many will tell you to do this after cleaning. I do it before cleaning because SFACL4 is a non-destructive editor for the most part and Normalization can be undone or modified and there is a second reason. Look up what Normalize does. Press N. I suggest normalizing to -1 dB. You can see what the program is proposing under Peak change. If you're getting a Peak change of more than +15, then you probably recorded too low.

After normalizing, the waveform will change accordingly. There is a middle handle that is the object volume. It should have moved up be the amount of volume increase. Click on it and hold the mouse button down. You can always adjust the object volume here.

A sudden loud noise, like a cough close by, can mess up normalization. I'll discuss problems like this later.

Now when you play back, the volume will either be louder or the same. And you will probably hear the background noise much louder. I find that it is better to get the noise removed/reduced when it's loud and not find out later that there is still unacceptable noise once the volume has been increased.

Next, Cleaning.

There is an Auto Cleaning tool at the top right of the Cleaning window. Try it to see what it does. Any of the Magix De- whatever tools that it uses will be turned on when you click on Cleaning. You can tweak them.

Alternatively, probably better, and as a comparison, try iZotope RX10 Repair Assistant. First, turn off any other cleaning effects. In the Repair Assistant interface, click on the ? and read how it works. In the help text, make sure to read the correct part as you are using the Repair Assistant not the Repair Assistant Module. Click on the Voice in the interface if not selected by default. Click on Learn. Start playback preferably where there is just background noise for a couple of seconds, then speech. It will listen and then create a suggestion. Stop playback. Look at what it did. There are 5 De's. DeHum is not one of them as that is just part of De-Noise. Play back and listen. Any better?

You can tweak the tools. You don't want to hear any artefacts or chirping. Play with the tools by turning them down, then up. Bypass. Sometimes, some background noise is desirable, so adjust the dials to suit.

As another test, you can play with the tools individually. Start with the obvious - De-noise.

This step should take care of most of your problems.

Next, Spectral editing

If there are any sudden noises, clicks, taps, door closing, a cough, etc., you can click on Spectral in the upper right menu and try removing these noises. Read up on how to do it. It takes practice to get it right.

Next, Volume adjustment - compression combined with volume curve adjustments.

Compression will make weaker speech closer to louder speech. Read up on it. I mentioned using the Melda Compressor because it has the traditional graph, making it easier to understand. More on this later.

Off for a drink by the fire.

John CB