Community of Sexsoundlovers

Please can anyone help? I have some old cassettes that were recorded at low levels and I seem to be going round in circles trying to clean them up. I wonder if anyone knows the answer to either of these questions. I'm using Adobe Audition, but think the tools will be common in most progs.

Question 1: The recordings have some peaks that are much louder than the sounds I'm interested in. The first thing I try to do is remove them and I want to do so without altering the normal room sounds. Assuming the loudest interesting sound I want to work on is -35 dB, if I use a Hard Limiter at the following settings will it just remove peaks above that volume and leave the lower volumes untouched? The settings I assume would achieve this are Maximum Amplitude -34 dB and Input Boost 0 dB.

Question 2: After removing peaks I want to Normalize, but also want to use EQ to remove some lower and higher end frequencies, plus doing some Noise Reduction. Does anyone know which is the best order to do these three things in? I'm starting to wonder if it's EQ, followed by NR and then Normalization. I say this, because if I Normalize first, both processes (EQ the way I use it and NR) do remove some volume from certain frequencies. On the other hand I find it harder to hear the effect the NR is having before normalizing as the original volume is low.

Any help would be much appreciated, as would suggestions for alternative ways of doing things.

This is all going to be quite subjective and I've been out of education for long enough to probably have some facts wrong, but perhaps it'll help.

I'd say that if those huge peaks are short sounds like doors closing or incidental sounds that are unlikely to be masking good content such as a toilet flushing, don't bother trying to tame them, just scan through the waveform and chop out any huge peaks of that type. Very distructive approach yes, but in my experience they can rarely be tamed down to the point where they're worth keeping. While you're at it, some drastic low cuts can be applied early on. Unless you're monitoring audio from through a wall or in some other unusual environment, there'll be nothing worth hearing below 100 Hz other than the rumble of passing cars/planes or the hum of fridges etc. Depending on the environment, I've shelved as high as 300 Hz before without losing anything specific to our little hobby.

As I understand it, maximum amplitude would be the maximum dB that level could reach, at which point the limiter is a brick wall that can't be exceeded. Keeping in mind that in music, limiters are generally a tool used as one of the final stages of mastering to smooth out the dynamic range and raise the overall level to broadcast standard, I'd say you're probably using it in the wrong context. If I wanted to use the limiter first, it'd be to raise the overall level so that I could get an idea of what material I'm working with and what EQ settings would be necessary, in this example I'd set maximum amplitude to about -10 dB and input boost to somewhere around 25 dB. As I understand it, the limiter would then go through sample by sample (44100 samples per second unless you're working at a higher bitrate), examine each sample and calculate what amount of dB needs to be added to bring your audio up to that maximum amplitude value of -10 dB, with a maximum possible 25 dB of gain that can be added (even if adding the full 25 dB means that some samples don't reach your maximum amplitude level). Assuming that the -35 dB was accurate, you'd have your loudest interesting sound raised up to -10 dB, with the dynamic range of quieter interesting sounds fairly well preserved in comparrison to that. The -10 dB is there as headroom, so that you can be extreme about boosting mids and tops for clarity if needs be without clipping. I'd use Normalize to make the final climb toward 0 dB if the waveform was consistent once I'd finished, but if the audio had a larger dynamic range that you didn't want to clamp down on using compression then you'd need to adjust the numbers because normalizing only applies to the highest peak, so you could end up with 1 second at 0 dB and almost everything else at -15 or something.

Question two leads me to wonder how Audition works (sorry, never used it). Is it a destructive or a non-destructive editor? I mean, once you've chosen your limiter settings, do you have to click apply and wait for it to process the whole file before moving on to the next stage? If that's the case, your workflow is going to be insanely long because every decision will take tons of number crunching with the length of files you're examining. A better approach would be a non-destructive DAW that applies the plugins in realtime, then you export or "bounce" a finished mix to disk once you've tweaked the plugins to taste. For a good free option, check out Reaper from www.reaper.fm. There are a zillion ways to skin every cat in that DAW, so can be intimedating, but the stock compression limiting and EQ plugins are pretty good and the forum is full of helpful chaps.

Only other tip I'd say is that IMHO noise reduction should come last and be used sparingly, because unless you're going to invest serious money into something pro like iZotope RX and time to get good at using it, more often than not it ends up degrading the environment to my ears.

The second half of this is useless to you of course if Audition turns out to be non-destructive already, but I seem to recall it being aimed at post-pro and radio types where destructive is the norm, so thought it was worth asking.

Does any of that help/make sense?

Sc0tt-uk

sc0tt-uk wrote:
Does any of that help/make sense?

Sc0tt-uk

Thanks a bundle for taking the time to reply and it seems to make a lot of sense. I do need to read it properly and try some things out, but will reply fully either later this evening, or tomorrow. I have been doing a bit of experimenting, so will bring you up to speed with that to, when I do respond. I really appreciate the feedback.

Thanks again for taking the time to reply and the useful information you provided.

sc0tt-uk wrote:
Question two leads me to wonder how Audition works (sorry, never used it). Is it a destructive or a non-destructive editor? I mean, once you've chosen your limiter settings, do you have to click apply and wait for it to process the whole file before moving on to the next stage? If that's the case, your workflow is going to be insanely long because every decision will take tons of number crunching with the length of files you're examining.

I'll start by clearing up issues about Audition. It is non-destructive while working on the recdording, but if you save in Waveform, which I use, it is destructive. It's easy to get round that though, as I use 'Save As' in a new project folder with something like 'V2' as soon as I open a file. I sometimes save new versions after each process and clear up from time to time.

On the subject of speed it's not an issue, as the computers primary use is video editing, so by comparison audio is lightning. As an example, if I want to normalize 45 mins it takes about 15 secs or less, so undo and redo are easy. It's probably the same in other progs, but it also opens each effect with your last used settings.

I really like Audition, but don't have much choice anyway, as I use it with Premiere Pro and they work seamlessly together. Another little fact that might be of interest if you use Photoshop, is that Audition has few tools similar to that. A Marquee And Healing Tool for instance.

I'd say that if those huge peaks are short sounds like doors closing or incidental sounds that are unlikely to be masking good content such as a toilet flushing, don't bother trying to tame them, just scan through the waveform and chop out any huge peaks of that type. Very distructive approach yes, but in my experience they can rarely be tamed down to the point where they're worth keeping.

I totally agree with you here, but it's the time factor. With my old cassettes the interesting sounds are all quite low volume. On a tape I tried removing peaks from recently, the loudest sound of interest was -39 dB, so the number of peaks to remove was pretty massive. I did try but it took an age. That's why I went back to looking at Hard Limiting, to do it for me. Again, with a quick undo/redo facility it doesn't take time to try different settings and listen on key sections.

While you're at it, some drastic low cuts can be applied early on. Unless you're monitoring audio from through a wall or in some other unusual environment, there'll be nothing worth hearing below 100 Hz other than the rumble of passing cars/planes or the hum of fridges etc. Depending on the environment, I've shelved as high as 300 Hz before without losing anything specific to our little hobby.

I agree with you here and use EQ after removing peaks. On the low end I tend to use 100 Hz and below at zero, 125 Hz at -10 DB and the rest at -0 dB. Having noted your figure of 300 Hz I think I might be being a bit conservative for our purposes. I also cut some of the top end too and tend to do it with 10k at -0 dB, 12.5k at -10dB and all above at zero. If you have any opinions on the sense or lack of it with this, do let me know.

As I understand it, maximum amplitude would be the maximum dB that level could reach, at which point the limiter is a brick wall that can't be exceeded. Keeping in mind that in music, limiters are generally a tool used as one of the final stages of mastering to smooth out the dynamic range and raise the overall level to broadcast standard, I'd say you're probably using it in the wrong context. If I wanted to use the limiter first, it'd be to raise the overall level so that I could get an idea of what material I'm working with and what EQ settings would be necessary, in this example I'd set maximum amplitude to about -10 dB and input boost to somewhere around 25 dB. As I understand it, the limiter would then go through sample by sample (44100 samples per second unless you're working at a higher bitrate), examine each sample and calculate what amount of dB needs to be added to bring your audio up to that maximum amplitude value of -10 dB, with a maximum possible 25 dB of gain that can be added (even if adding the full 25 dB means that some samples don't reach your maximum amplitude level). Assuming that the -35 dB was accurate, you'd have your loudest interesting sound raised up to -10 dB, with the dynamic range of quieter interesting sounds fairly well preserved in comparrison to that. The -10 dB is there as headroom, so that you can be extreme about boosting mids and tops for clarity if needs be without clipping. I'd use Normalize to make the final climb toward 0 dB if the waveform was consistent once I'd finished, but if the audio had a larger dynamic range that you didn't want to clamp down on using compression then you'd need to adjust the numbers because normalizing only applies to the highest peak, so you could end up with 1 second at 0 dB and almost everything else at -15 or something.

Thanks so much, that's a great explanation. It took a bit to get my head round it, but it's helped me work out what I've been doing wrong. While I was looking at it, I tried some of the settings you suggest and also played with altering them. I'm now pretty sure I can use the Hard Limiter to remove any sound above a certain volume, while not altering anything below that volume. The answer was to set the input boost at 0.0 dB. I hadn't noticed before and the default gave the lower volumes a boost. I tried it on a bit of the recording with -39 dB as the highest useful volume. I found I could even push it to a Max amplitude of -55 dB without any significant difference to the interesting sounds. I know it worked, because I could hear sounds I hadn't heard before and I could also recognise some sounds that had been blurry, because they were masked by louder sounds. I tried the HL on some other bits and got similar results, but wasn't able to push the Max Amp much past the loudest useful bit on those.

The bottom line though, is I'm pretty sure I can use the HL as a quick method of removing anything above a certain volume and keeping everything below it by setting Input Boost to zero. I happily stand to be corrected on this if it's my imagination doing the listening, but it does seem to work.

check out Reaper from http://www.reaper.fm. There are a zillion ways to skin every cat in that DAW, so can be intimedating, but the stock compression limiting and EQ plugins are pretty good and the forum is full of helpful chaps.

Thanks for the link to the forum, much appreciated!

Only other tip I'd say is that IMHO noise reduction should come last and be used sparingly, because unless you're going to invest serious money into something pro like iZotope RX and time to get good at using it, more often than not it ends up degrading the environment to my ears.

I agree with what you say about the degradation. Audition isn't too bad as long as you don't expect miracles. I did use it on the Lynn recording I did and it made a bit of difference. In my opinion, it's worth remembering it's reduction and not noise elimination!

Sc0tt-uk

Thanks again Sc0tt-uk. Until I get corrected, or learn more sophisticated methods, I'm going to try the following in this order.

Hard Limiting with Maximum Amplitude set around the loudest interesting sound and Input Boost set to zero.

EQ as described above but maybe reducing more lower frquencies.

Normalize at -6 dB to leave the headroom Sc0tt-uk suggested.

Noise reduction but not annihilation.

Tweaks in Spectral Frequency Display.

As Sc0tt-uk said, I hope that makes some sense.

Apologies for using the quote marks wrong in the previous post. If anyone wants to point out to a simple soul how I should have done it, then please go ahead. I will try harder in future.

Just checking in quickly...

What the hell are your computer specs... 15 secs to Normalize 45 mins of audio is insane! (pats his battered old Mac to try and ease its insecurities LOL)

Well, set like that, the HL would be performing extreme gain reduction on those peaks, as well as gain reduction on the louder interesting sounds, taking them all down to the lower level which I guess is the same as unmasking those lower level sounds when the chips are down. If Audition's HL can do that without leaving a jumbled mess behind, hats off to Adobe! Next time I'm tidying up a catch I'll have to give this a go, sounds like it could be a real timesaver.

Yep, that EQ shelving on the highs makes sense here when you're transfering from cassette. Seems like you're letting your ears be the judge rather than totally rlying on what the audio looks like anyway, so you can't really go wrong

Really looking forward to checking out your work man!

Sc0tt-UK

sc0tt-uk wrote:Just checking in quickly...

What the hell are your computer specs... 15 secs to Normalize 45 mins of audio is insane! (pats his battered old Mac to try and ease its insecurities LOL)

Sc0tt-UK

I started to think I was being daft by saying 15 secs when I saw your reply, so have just checked it out on a 45 min 925 MB recording. I was wrong it's more like 11 secs!

The computer is needed for video editing, so it's either making lots and lots of cups of tea, or raw processing power. I had a multi-cam project last Autumn, so decided to upgrade. I can't afford the sophistication of a Mac, so got a Bare Bones Bundle from Novatech. That's case, power supply, motherboard, processor and RAM. The essential figures are an Intel i7 2600K quad core processor, with 8 GB of DDR3 1600 Mhz RAM.

Out of interest, I just checked the Novatech Website and you can get a more powerful one now, that's less than what I paid for mine! It's their second to top model and all you would have to do is replace the RAM. it would be less than £450, but I know it's a lot of money.

I am working on a recording at the moment and hope to post it next week, but have to check something out with yeppie first. It's from a very low level cassette and I'll post a bit of the original as well, so you can see what you think to the process.