Help with AVI File

I am needing assistance with trying to obtain the background audio on a AVI file. There are words that I am needing to find out clearly and need a sound expert to direct me the best route for doing so with software or who I can speak with that could professional assist me.

 

What are you trying to do? I dont understand what you mean by obtain?

 

I have a audio recording from a digital camera that has a person speaking in the background of which is what I am trying to obtain the words from. I need to find out how to eliminate the other noise in order to hear correctly what this person is saying. Do you have a direction of where I can go or who can assist me with this?

Thank you in advance for your help

 

I short the answer is no. you can use filters to help improve the clarity by removing frequencies above and below the voice and parametric eqs to reduce narrow band frequencies but the filters will also remove the frequency from the vocal so you may get un-natural sounding vocals.

You can also use destructive interference but you have to have an exact copy of the background noise to invert to leave just the vocal. If you have recorded the audio from a camcorder mic you wont be able to use this method.

If it isnt recorded good first time round then it is difficult to improve it. The films lie!

 

Hi Tueur,
Would Audacity at least give the opportunity to try to clean things up just enough to decipher the words? I don't do any audio work but it seems Audacity might be able to alter the sound enough to possibly pick out words if that is steelhead's intent. I may be completely off on this which is why I am asking.

 

Audacity does have some basic filtering options but the main problem is that the human voice has a very wide frequency range (500Hz to 12kHz I would estimate) This means that any filterin will inevitably affect the tonal quality of the vocal. It may improve its audability but it wont be very natural sounding. Depends on what you want to get out at the end. ie if your spying on someone and want to know what they are saying you probably dont care about the tonal quality, however if you are trying to improve the audio on a home video it is a bit more tricky.

 

Thank you both for the feedback and direction. I am actually just trying to pick up the words and don't care about the tone of the voice. Its a recording that I am just ultimately trying to find out the exact sentence or words spoken. In the forward sound is snoring close to the mic so that is what I am needing to eliminate and pick up the person speaking in the back. Do you know of anyone that may do this as a profession that could assist?

 

I think what you are after is impossible. The snoring will be in the same frequency range as vocal so anything that you do to reduce the snoring will also reduce the vocal.

You see people enhancing audio and photographs in the film all the time but in reality it is not possible. Data , whether it is audio, video or whatever, is only as good as the point that it is captured. If your raw audio is "missing" information you cant create it. That is why you always capture at full quality the compress after. The only analagy I can think of is that it would be like an author writing a book using only nouns and the computer enhances the story to include all the words.

If you have a regular niose like mains hum that you can replicate you can sometimes use destructive interference to clean up audio but due to the random nature of the peaks and troughs in the snoring you wont be able to apply this technique to your work.

Have a read about these noise cancelling headphones which use the principles of distructive interference to cancel background noise. this would be the only way that you could get the result that you want but without a an exact copy of the background noise you cant apply it.

http://electronics.howstuffworks.com/gadgets/audio-music/noise-canceling-headphone3.htm

If the person speaking is in the video then you might be able to find someone who can lip read.

 

I should have said that the only time i have tried this technique to extract the vocal from an audio track. If you get the vocal version then get the instrumental version and invert it the overlay it the instrumental track cancels out the instruments in the vocal mix leaving only the vocal.

The only caveat is that the instrument has to be the same recording in both instrumental and vocal mixes... ie the vocal has been dubbed over after