Compare Audio Files - Extract Common Sound

[ContactGT]

I have 2 videos of wedding speeches. One shot on my HDV Canon HV20 and another shot on a DV Sony 'something'.

I have both versions on my computer and am editing the video in Premier CS4.

Is there some software that can examine the 2 audio files and pull out the common audio, so that I can get decent quality version of the speech. Both audio files are of the same source, but 1 is from the back of the room, so the speech is quiet and the surrounding audience noise is loud. The other was shot near the front of the room and has good audio of the speeches, but lots of cuttlery and glasses 'clinking' and people mumbling round the table. Neither track is particularly good, but between them I think the information must be in there.

Can I feed these 2 files into something and produce an audio file consisting of just the common sound?

[t0nee1]

Try an audio editor like Audacity...

[ContactGT]

I have looked into various audio editors. I have SoundForge - it can merge tracks etc, but I can't find a "compare and find common" feature. I did try to google for this but found (amongst the rubbish) several freeware tools that will compare tracks, but they sound like they only work with decent quality originals and just compare pre and post transmission and remove crackles and pops - I have 2 totally different sounding tracks, but both have the same underlying voice track at different volumes and quality.

I have just read through the blurb for Audacity and, I might have missed something, but at a glance it doesn't sound like it does anything more than soundforge.

[greymalkin]

even if you were to find something like that the result would not be good, you can't make a good audio track from 2 bad ones.

the only way to get good audio for situations like that is to have a seperate microphone close to the speaker.

pick the least offensive track and clean it up the best you can.

[ContactGT]

OK. Thanks. The 2 audio tracks are not bad quality (as in noise and interference), they just both have other sounds that occasionally block out the speaker's voice. Between them, they have the full speech - I'll see what I can do in soundForge - a clever bit of editing I think...

GT

[guns1inger]

This is the sort of job I would do (and have done) in Vegas. I align the two tracks so they play in sync, then use the volume envelope to mix and mute the different parts of the two tracks to create a single track with the best parts of each. While it is still a manual process, once you get going it is relatively quick.

[poisondeathray]

I'm not sure if soundforge has this, but audition & soundbooth have a spectral view where you can manually select sections to delete or keep. e.g. if there are glasses clinking or cellphones ringing, the freqencies are usually very dintinct from , say, the speech or background music, and this is visually displayed. This too is a manual process, but the "occasional" nature requires it to be (as opposed to regular noise removal techniques)

[newuser2009]

Hi,

I would recommend to try this software:

spam

I am sure the trial version one can download from the site will work just fine for you.

And by the way, if anybody else evaluated this software I would be happy to get feedback... our company is considering buying it for HD Voice testing in our VoIP environment. Thanks a lot in advance!

You are in breach of the forum rules and are being issued with a formal warning.
/ Moderator Baldrick

[filmboss80]

First of all, you can equalize and adjust levels in each audio track (using Audacity) until each sounds close to the other in tonal quality.

Then lay each audio track in the Premiere timeline and synchronize them at various points using Premiere's marker feature. When switching audio sources, fade out one as you fade in the other.

This sort of thing is done all the time, and if you familiarize yourself with your editing software, you will find it's not too hard to accomplish.