Forensic style speech recovery from cassettes of sermons

morning -

i have an assortment of cassettes of church sermons that i am archiving. some are recorded so low, any gain just raises the noise floor. i use normalize and level 2 voice compress (plus noise reduction) but it is difficult.

any thoughts? also, what is the best plugin to average (apply an agc of sorts) to a speech recording as the pastor starts off low and then gets louder.

thanks ya’ll. :slight_smile:

Steve’s AGC plugin (from 2012) does work in Audcaity3 …

Steves AGC plugin still works

AGC - Automatic Gain Control #

thanks - i’ll hafta snag that up - and redo 200+ audio tracks :frowning: maybe.

Audacity can do batch-processing …

thanks :slight_smile: but i cannot locate agc v 2 and all pages to it are no longer avail. :frowning: and when i used the default voice, it really raised the noise level in a bad way.

Turning up the gain will inevitably raise the noise with the voice.

AI may help isolate the speech …
Enhance Speech from Adobe | Free AI filter for cleaning up spoken audio [Beta & free]

[ NB: The speech recovery shown in movies is science-fiction ].

Do you do noise reduction before you apply compressor with gain? Our church records sermons on CDs and the volume is rather low. I use the regular compressor with the following settings:

Threshold: -30 db
Noise Floor: - 40 db
Ratio: 8.5:1
Attack time: 0.10 secs
Release time: 2.0 secs

I have checked “Make up gain for 0 dB after compressing” and also “Compress based on peaks.”

Then I run noise reduction again. I used to run compressor first, but I finally figured out that during times when there is extra noise it’s better to run noise reduction FIRST. Then there’s not as much remaining noise which gets enhanced by the compressor/gain. My biggest issue is that loud breaths and other noises are also enhanced by the compressor, and I like to listen to our sermons and correct as much as I can manually. I can’t imagine trying to manually edit an a large number of them, though. We previously used cassettes, and I used a similar setting for them but it was in a much older edition of Audacity. I don’t think there was much difference between the noise level of our cassettes vs the CDs. It’s just easier to “rip” the CDs onto my laptop whereas I used to have to plug a cassette deck into my laptop and actually play the cassette and re-record through Audacity.

I have tried the Adobe AI thing, but it did not raise the volume very much and when I then ran compressor I got similar results to what I get without using the AI filter. The AI filter also took out the music that often played at the end of the sermon when the pastor was transitioning from the message to the invitation, and I didn’t like that.

:smiley: that was perfect on two tapes. and i’m just about done over 150 tapes… now i need to revisit some of my problems and redo them. i do one more effect - notch filter for the occasional 59-62hz hum. :slight_smile:

these are tapes from 2001 - 2006 and not recorded under the best of conditions. nor stored in same… :frowning:

i have use a tascam 424 to capture both sides at dbl speed - then save the capture as a project, then try and work magic. so looks, like i’ll have some work to do - but the ‘heavy lifting’ has been done. the only caveat, the audacity file does not have any info beyond date (no verse/etc).

i just redid a ‘problem - challenging tape’ and this method worked well. thanks :slight_smile:

This topic was automatically closed after 30 days. New replies are no longer allowed.