I''m creating a live consert DVD for some friends'' of mine''s band. Unfortunately we made a screw up with the CD-Recorder so the CD is useless. Fortunately I got everything on video and I''m going to use that for the sound (I had one camera on a gallery in the consert hall). I had some problems with that as well, but it''s the best source I''ve got: it recorded in 32Khz instead of 48 or 44.1 so it doesn''t sound too good, and it had an AGC. The result is that I have no dynamic and when the vocalist sings everything else gets lower. (The same with bass vs guitars, drums vs everything etc) I''ve just spent 7 hours or so manually editing the volume envelope for it with Adobe Audition.. I managed to get it to sound more or less ok, but is there a better way to do it? I''ll never want to do *that* again Another problem altogether is that it sounds like the speakers are inside a cardboard box - the accoustics of the consert hall didn''t exactly help me get get a good recording.. I''ve tried fixing it with the EQ, but there''s some weird reverb and probably phase problems as well. Any idea how I can get it to sound better? I''d post a sample, but I''ll have to get permission from the band first and I won''t see them until monday..

Well, you can''t make something out of nothing. You might be able to restore some of the higher frequencies to the sound with an ''exciter'' filter. I doubt there''s anything you can do about the reverb if that came from the venue. Maybe a dedicated unidirectional microphone would help there. Can you turn the AGC (automatic gain compensation? I dunno) off in future? Or set the recording level lower? Try and do some recording during the soundcheck to get an idea of a good recording level and possibly find the best place to be positioned in the venue.

[ MSVC Fixes | STL Docs | SDL | Game AI | Sockets | C++ Faq Lite | Boost
Asking Questions | Organising code files | My stuff | Tiny XML | STLPort]

It was a one shot go.. I don''t think they''ve got any more concerts planned in the near future...

The camera wasn''t mine and I wasn''t familiar with it... I don''t know if it''s possible to turn off the AGC (Automatic Gain Control -- extreme compression). I should have noticed the 36KHz setting though.. It would have been better with a full frequency range, but that''s not the biggest problem.

The CD-Burner was connected to the matrix of the desk. In theory it would work perfect, but we didn''t connect it until after the soundcheck (not my idea).. Stupid, I know I did try to get a decent sound for the output though, but I only had a lousy headphones with only one working "speaker".. Vocals and bass are ok, but the drums and guitar is a bit low.. I also forgot to add the effects to the mix so everything is dry.. It sounded good on the headphones but then I probably heard more from the PA than the headphones . The recording sounds bad and because nobody had told the sound engineer he had to turn it on so we lost the first 15 minutes or so. And to add insult to injury the CD-Recorder automatically shut down during the concert because of a too long period of silence so we lost the last 15 minutes as well

The probably best solution is to use a "studio" recording they made earlier and match the tempo.. Some creative view shifting should hide most if not all the sync problems. - We used three cameras - one stationary and two roaming (ie worse sound than the stationary). The video looks good, and that was what they really wanted me to do.. The sound thing wasn''t really my responsibility

What I''d really like is a filter or software that can generate a volume envelope based on the amount of background noise/hiss that the camera picked up.. The AGC cranks up the gain if the volume is low (much hiss) and turns it down if the volume is high (less or no hiss). If I could somehow get a curve based on this I could probably rectify the problem.. I did it by ear the last time and it took about 7 or 8 hours to get it to sound decent, but I''m not happy about it yet so I may be looking forward to more hours of moving tiny points on the envelope..


On a side note:
I got a possibly excellent idea as I was driving home from work listening to the tape.. If it were possible to treat the spectral view in a sound editor as a photoshop image then it would be easy to "darken" to make an area (time*frequency) have a lower volume and "lighten" to raise the volume. Various pencil and brush tools could be used to "paint" in new frequencies or remove existing frequencies. I don''t know if it''s possible though.. If the spectral view is "one" way then it won''t work - except maybe as a control for various filters. "Darken/Lighten" would corrospond to an EQ setting for that area and time and drawing a frequency would corrospond to generate tones.
Each pixel of the spectral view would ideally corrospond to a band of 1 or less Hz for a duration of 1 sample.

1 second of 44,1KHz would be at least 44100x22050 pixels at maximum zoom though - so I don''t think it would be easy
A control for a dynamic/parametric EQ and tone generator should be possible though.