12-13-2009, 12:08 AM
Questioner,
You pretty much ran down the same list I did in terms of trying to clean up the audio. I have a pretty extensive VST library, which includes some noise canceling plug-ins. Since analog tapes are one of the noisiest mediums to clean up audio from, the methods you described are great in theory, but have some crazy side effects when tweaking them.
Gary was my second set of ears when he was here in Washington DC visiting me, and he can attest to the difficulties when trying to optimize old analog audio. I've tried high/low pass filters, isolated equalization, band compression on only the voice frequency ranges, and noise gates with varying degrees of attack and release.
The main problem was that the audio quickly began to sound fake. Gates become completely distracting to the ear, because no matter what degree of gating you do, the ear tends to zero in on the "hiss-quiet-hiss-quiet" effect rather than being able to concentrate on the words. Equalization takes away the brightness of Carla's voice and brings it down more towards telephone frequencies while still preserving a degree of dampened hiss. So that was definitely a no go.
Noise "learning" has been the best approach, by far. By using BIAS's SoundSoap Pro 2, we can sample a "quiet" section of tape hiss and ambient room noise, and then diminish those exact frequencies across the tract.
Since the application does a great job with batch processing against a custom profile of processing parameters, we can start cranking out clean audio automatically, and then compress it and normalize it to make it more comfortable to listen to.
Not only will we clean up the Ra sessions, but we'll also be able to do the same to the Q'uo sessions, as well. So stay tuned... :-)
Steve
You pretty much ran down the same list I did in terms of trying to clean up the audio. I have a pretty extensive VST library, which includes some noise canceling plug-ins. Since analog tapes are one of the noisiest mediums to clean up audio from, the methods you described are great in theory, but have some crazy side effects when tweaking them.
Gary was my second set of ears when he was here in Washington DC visiting me, and he can attest to the difficulties when trying to optimize old analog audio. I've tried high/low pass filters, isolated equalization, band compression on only the voice frequency ranges, and noise gates with varying degrees of attack and release.
The main problem was that the audio quickly began to sound fake. Gates become completely distracting to the ear, because no matter what degree of gating you do, the ear tends to zero in on the "hiss-quiet-hiss-quiet" effect rather than being able to concentrate on the words. Equalization takes away the brightness of Carla's voice and brings it down more towards telephone frequencies while still preserving a degree of dampened hiss. So that was definitely a no go.
Noise "learning" has been the best approach, by far. By using BIAS's SoundSoap Pro 2, we can sample a "quiet" section of tape hiss and ambient room noise, and then diminish those exact frequencies across the tract.
Since the application does a great job with batch processing against a custom profile of processing parameters, we can start cranking out clean audio automatically, and then compress it and normalize it to make it more comfortable to listen to.
Not only will we clean up the Ra sessions, but we'll also be able to do the same to the Q'uo sessions, as well. So stay tuned... :-)
Steve
(12-12-2009, 12:51 PM)Questioner Wrote: 1. Use a noise gate with a long release time and slow attack to reduce background noise between talking, while leaving in all the speech content.
2. Use a noise reduction filter that can sample background noise, identify the dominant frequencies and reduce them, again without garbling the speech.
3. Use a compressor to even out the volume of the speech, then peak normalize for a consistent level.
4. Discover equalization settings that include clarity for each speaker's voice. Split the audio into a track for each voice so that the EQ per speaker can be applied throughout the session.
5. Add an overall low frequency cut to reduce rumble and wind noise, and an overall high frequency cut to reduce tape hiss. Gate the high cut so that it is only present between times of talking.
In the next few weeks I might be able to participate in the audio editing project.