Feedback elimination? -Sound Forge - PC - MAGIX Community Forums Archive

Subject:Feedback elimination? Posted by: JazzFrets Date:10/15/2002 9:41:19 AM OK, I've got an acoustic guitar track (fingerstyle steel string) that is an almost perfect take -- except for about 2 seconds where there is a bit of ringing that is verging on feedback (odd, since there were no speakers in the studio -- it was a headphone mix). I don't want to scrap the track, so I used the SF spectral analyzer tool to try and find the frequency(s). There seem to be two different frequencies: 117hz and 282hz. I applied two narrow notches of SF parametric EQ at those frequencies (approx: -9.7db, .4 octive wide). It helps, but I'd like to do better without attenuating things that don't need it. Since the track is fingerstyle, there are some other notes happening that I don't want to mess up their EQ. Suggestions? Is there a direct x "feedback eliminator" plug-in that works something like an hardware unit that I might be able to use to do a better job?
Subject:RE: Feedback elimination? Reply by: Rednroll Date:10/16/2002 11:05:12 AM You can get feedback by the headphones bleading into your microphone if your mic gain is set very high, which it probably was to record an acoustic guitar. A hardware feed back elliminator has a "look ahead" function where it looks at the audio signal and then outputs it a few milliseconds later. It can detect feedback and then pinpoint where that frequency is, and it adjusts itself to reduce the frequency where the feedback is occuring, thus stopping the feedback from occuring. Once the feedback has occured and is recorded the only way you can reduce it, is too find the particular frequencies that the feedback occured and reduce that frequency. This is what you have already done. The only thing that a hardware feedback elliminator can do for you now is help find that particular frequency that the feedback occured at. Some will make the Q too wide, so it will effect a lot of other frequencies around it. So the way you're doing it, is probably the best way. Just apply the EQ to the sections that the signal started to feedback at, so you don't effect the entire guitar part and try to keep your Q as narrow as possible. If that isn't acceptable then go back and rerecord the guitar part and this time turn down your headphones or get a pair that has a better seal to avoid leakage problems.
Subject:RE: Feedback elimination? Reply by: JazzFrets Date:10/16/2002 11:43:11 AM Thanks for the feedback...uh, I mean no feedback...well, you know what I mean. I had not thought of the headphones as a problem. It's a pretty high end acoustic guitar with a ton of sustain and an internal condenser mic to boot. That's probably it.
Subject:RE: Feedback elimination? Reply by: Chienworks Date:10/16/2002 12:33:40 PM I would be a bit suspicious of the frequencies reported by spectrum analysis. It seems to be a bit off and buggy. I just tried generating a 282Hz sine wave, then running spectrum analysis on it and it reported 280Hz. A 117Hz generated wave showed as 108Hz. If your analysis is off that far, then you might be missing the frequency. You should try playing with the center frequency of the notch filter a bit and see if that helps.

Subject:Feedback elimination?

Posted by: JazzFrets

Date:10/15/2002 9:41:19 AM

OK, I've got an acoustic guitar track (fingerstyle steel string) that is an almost perfect take -- except for about 2 seconds where there is a bit of ringing that is verging on feedback (odd, since there were no speakers in the studio -- it was a headphone mix). I don't want to scrap the track, so I used the SF spectral analyzer tool to try and find the frequency(s). There seem to be two different frequencies: 117hz and 282hz. I applied two narrow notches of SF parametric EQ at those frequencies (approx: -9.7db, .4 octive wide). It helps, but I'd like to do better without attenuating things that don't need it. Since the track is fingerstyle, there are some other notes happening that I don't want to mess up their EQ. Suggestions? Is there a direct x "feedback eliminator" plug-in that works something like an hardware unit that I might be able to use to do a better job?

Subject:RE: Feedback elimination?

Reply by: Rednroll

Date:10/16/2002 11:05:12 AM

You can get feedback by the headphones bleading into your microphone if your mic gain is set very high, which it probably was to record an acoustic guitar. A hardware feed back elliminator has a "look ahead" function where it looks at the audio signal and then outputs it a few milliseconds later. It can detect feedback and then pinpoint where that frequency is, and it adjusts itself to reduce the frequency where the feedback is occuring, thus stopping the feedback from occuring. Once the feedback has occured and is recorded the only way you can reduce it, is too find the particular frequencies that the feedback occured and reduce that frequency. This is what you have already done. The only thing that a hardware feedback elliminator can do for you now is help find that particular frequency that the feedback occured at. Some will make the Q too wide, so it will effect a lot of other frequencies around it. So the way you're doing it, is probably the best way. Just apply the EQ to the sections that the signal started to feedback at, so you don't effect the entire guitar part and try to keep your Q as narrow as possible. If that isn't acceptable then go back and rerecord the guitar part and this time turn down your headphones or get a pair that has a better seal to avoid leakage problems.

Subject:RE: Feedback elimination?

Reply by: JazzFrets

Date:10/16/2002 11:43:11 AM

Thanks for the feedback...uh, I mean no feedback...well, you know what I mean. I had not thought of the headphones as a problem. It's a pretty high end acoustic guitar with a ton of sustain and an internal condenser mic to boot. That's probably it.

Subject:RE: Feedback elimination?

Reply by: Chienworks

Date:10/16/2002 12:33:40 PM

I would be a bit suspicious of the frequencies reported by spectrum analysis. It seems to be a bit off and buggy. I just tried generating a 282Hz sine wave, then running spectrum analysis on it and it reported 280Hz. A 117Hz generated wave showed as 108Hz. If your analysis is off that far, then you might be missing the frequency. You should try playing with the center frequency of the notch filter a bit and see if that helps.