Removing talking from an audio file. Is it possible?

Advertise with us
Find deals on ebay
Amazon
QuadraphonicQuad

Help Support QuadraphonicQuad:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
JonUrban

JonUrban

Forum Curmudgeon
Staff member
Admin
Moderator
Since 2002/2003
Joined
Mar 2, 2002
Messages
17,727
Location
Connecticut
I am working on a "pet project" that I finally have found time to do.

I have taken the soundtrack from the very rare (you try and find one) "One Trick Pony" late issue LaserDisc with Dolby Surround and am working on putting together a DVD-A for the car. Many think this movie and music are bad, but it really hits a note with me! ;)

The surround mix is quite good - it's not super discrete, but the vocals are center channel and the music beds have a nice space to them in the 4 corner speakers.

Anyway, there are a few complete performances and others with dialog over them. For example, at the end of the movie, the tune "Jonah" is playing in a nice surround mix as Jonah returns to the study to steal his master tape. At one point, he asks the receptionist if he can go back into the studio to get his glasses. This is the type of dialog I would like to eliminate.

Is there a plug-in for Sound Forge or AA that can kill the dialog but not the music? Something that's "hobby level", I don't want to spend a fortune on this.

Anyone have a hint?

THANKS
 
If you're using Adobe Audition, you can try the "Remove Center Channel" option or Vocal Remover. But they only work well if the vocals are dead center.
 
Similarly, Plogue ($79) has Bidules that can put the vocals to the center channel which can then be discarded. A couple passes, and I think you could eliminate the vocals from the R and L channels.

Hope this helps.
 
JonUrban said:
..."One Trick Pony" late issue LaserDisc with Dolby Surround and am working on putting together a DVD-A for the car...Is there a plug-in for Sound Forge or AA that can kill the dialog but not the music? Something that's "hobby level", I don't want to spend a fortune on this.
Anyone have a hint? THANKS
Click to expand...

The soundtrack LP, reel or CD was never released? A lot of the time the QS encoding is preserved.

However, if the dialogue as you say is fully centered, mostly when people score music for under dialogue, there's not a lot of center-channel scoring for that section anyway, and if there is, it's generally NOT within the same frequency band as the actor(s) speaking at the time. So AA Center Channel Extractor should do alright, but use the one from the top menu not the one from the side menu because the one from the side menu leaves the BG in Mono. Then you can EQ out any remaining dialogue if you have to.

But you might not even have to do that. I seem to recall One Trick Pony coming out with discrete M&E's on the analogue tracks as an LD Box Set in a Director's Commentary version in some country or other. Or it might be a SqueezeLD or some such, or PALPlus for Europe. If so, you might not have to bother w dialogue extraction on your commercial edition since in most Director's Commentary editions, the M&E (music & effects) tracks is isolated on the Analogue tracks in the LaserDisc. So probably all you'd have to tolerate is maybe footsteps or doors creaking or whatever. But in a movie that old maybe the FX track is centered too. Then you could just run CC extractor and have your mix that way.
 
ndiamone said:
The soundtrack LP, reel or CD was never released? A lot of the time the QS encoding is preserved
Click to expand...

The soundtrack CD has different recordings of many of the tunes on this release. I had never heard of a deluxe box set of this movie, which was considered to be a flop and a hit to Simon's career. (Graceland bailed him out)

This WB LD came out right at the end of the LD era, and as far as I know, is the only way to get the movie with a surround soundtrack and in widescreen. There were very few made.

The vocal extractor is an interesting idea, however the dialog is in the center channel, which is mono. Not sure it will work on a mono file.
 
Is there music in the center channel as well as vocals? Are you trying to remove the vocals and keep the center channel music?

I may have misunderstood the challenge.

George
 
Hi George,

Here's the deal. The song is playing, all 5.1 channels, in a section where there are no song vocals, the character in the movie speaks a sentance or two over the music as the song continues. It's this dialog I would like to dump keeping the music behind it.

I'll post a clip when I get to my other PC.
 
Lucanu said:
Ok, let's see an experiment:
can you physically unplug your center channel when you hear dialogs?
if actor voice is removed you're in the game with a center channel extractor for sure.
Click to expand...

He means as in actually perform physical task detach playback wire from amp to center channel vs. perform electronically to see what remain.

The source Dolby Matrix I assume. If this is case, simply highlight few music bar containing dialogue and perform Extract Center Channel.
Listen to result closely on headphones and see what frequency if any, dialogue overlap over score. Inside Extract Center Channel is Frequency Band Select. Choose. Maybe takes a couple passes.

EASY WAY:
First pass choose Leave Bass. Next Choose `Custom'. Remember vocal band frequency from previous test. Type in frequency band From-To. Perform Extract again with new parameters. Save result.

HARDER WAY More Effective:
If source DTS or AC3 LaserDisc then yes perform demodulate into 5.1 and record all 5.1 channels to hard drive first.

1. If listening in computer to JUST demodulated center channel, generally dialogue not in same frequency range as whatever scoring might be there.

2. So discrete center channel can be then played in computer, listening closely to find out what frequency band is dialogue and how much dialogue frequency band overlap with scoring. Then for few bars of music that contain dialogue, open Equalization, highlight those bar in the waveform, loop-playback and slide around EQ to find out what frequency band is dialogue and how much frequency band overlap with scoring.

3. See how close separation can be between dialogue for ONLY those FEW MUSIC BAR with dialogue and then leave the rest by itself. Perform erase frequency and see what remain. Save file. If not satisfied, repeat perform and save file. Never know which practice run end up good so save all trials.

Finish place new center channel choose from one or more trials edit together, final save, reincorporate back into 5.1 mix, print.
 
JonUrban said:
This WB LD came out right at the end of the LD era, and as far as I know, is the only way to get the movie with a surround soundtrack and in widescreen.
Click to expand...

Check International Laserdisc Database online anyway. For extracting audio, doesn't matter if picture is pan and scan. Maybe Japan, Europe or Hong Kong has discrete M&E version, pan and scan or not. Lots of odd items released in foreign countries. If PAL or PALPlus edition find, players easy to find on eBay. Or maybe person previously perform dialogue extraction and post online.

Good luck.
 
OK, here's the sample(s) so you guys can hear what I mean.

First, this is a 16bit/44.1 6 Channel wav file FILE

This is a DTS of the same section FILE

The above files are not meant to be opened in Media Player (I don't think MP recognizes a 6 channel wav file), so you might get a windows error.
You can be assured that these are virus free.

NOTE: This is only a 14 second sample for evaluation and should not infringe on anyone's copyright. I will delete these files in a few days.
 
ndiamone said:
He means as in actually perform physical task detach playback wire from amp to center channel vs. perform electronically to see what remain.

The source Dolby Matrix I assume. If this is case, simply highlight few music bar containing dialogue and perform Extract Center Channel.
Listen to result closely on headphones and see what frequency if any, dialogue overlap over score. Inside Extract Center Channel is Frequency Band Select. Choose. Maybe takes a couple passes.

EASY WAY:
First pass choose Leave Bass. Next Choose `Custom'. Remember vocal band frequency from previous test. Type in frequency band From-To. Perform Extract again with new parameters. Save result.

HARDER WAY More Effective:
If source DTS or AC3 LaserDisc then yes perform demodulate into 5.1 and record all 5.1 channels to hard drive first.

1. If listening in computer to JUST demodulated center channel, generally dialogue not in same frequency range as whatever scoring might be there.

2. So discrete center channel can be then played in computer, listening closely to find out what frequency band is dialogue and how much dialogue frequency band overlap with scoring. Then for few bars of music that contain dialogue, open Equalization, highlight those bar in the waveform, loop-playback and slide around EQ to find out what frequency band is dialogue and how much frequency band overlap with scoring.

3. See how close separation can be between dialogue for ONLY those FEW MUSIC BAR with dialogue and then leave the rest by itself. Perform erase frequency and see what remain. Save file. If not satisfied, repeat perform and save file. Never know which practice run end up good so save all trials.

Finish place new center channel choose from one or more trials edit together, final save, reincorporate back into 5.1 mix, print.
Click to expand...

THANKS! I'll give this a shot! (y)
 
Thanks for the sample Jon. Dialogue over vocals in all channels except the sub. Good puzzle.

The librarian was easy to eliminate. I saved the EQ setting for her.

The kid though, I've done the reverse. I can eliminate the music and hear only the kid. I haven't eliminated the kid to hear only the music. Seems like the reverse algorithm should work but haven't discovered it yet.

The July 4th margaritas may slow me down a bit. We may want to make this type exercise a regular feature of QQ Forum.

Love the tune by the way. Can't wait to hear the whole tune, or soundtrack for that matter.

George
 
ChristopherLees said:
listened to the same track from that amazon link..it's in a different key,( which could be the amazon sample being done incorrectly)..
Click to expand...

Hi Christopher, the song is in tune, I've found another sample and it plays fine and in tune.
Now the only difference is that the sample Jon insert has a kind of extra reverberation which it seems not to be present in the cd version.
BUT, here's the trick:
Jon can extract a center channel from the cd version and put the mono wave file in the 5.1 mix in the center channel.
Only thing is that he has to get in sync with his 5.1 mix.
Could this be done easily? :rolleyes:
 
You must log in or register to reply here.

Similar threads

kfbkfb
Small Talk - Promoting Surround Sound
Replies
0
Views
148
kfbkfb
kfbkfb
M
Listening to 5.1 files with Apple iPhone spatial renderer (and problems with Apple Music match)
Replies
3
Views
457
MartinGr
M
HenshawAC
Greeting from Prague // Pozdravy z Prahy!
Replies
0
Views
181
HenshawAC
HenshawAC
T
New user
Replies
1
Views
282
par4ken
par4ken
ConfusedCroat
Big hello from Croatia
Replies
1
Views
506
LB-V
LB-V

Latest posts

Back
Top