better auto region for speech -Sound Forge - PC - MAGIX Community Forums Archive

Subject:better auto region for speech Posted by: David Thiel Date:3/23/2010 4:06:41 PM The auto region tool isn't enough to automatically create clean regions for extracting speech clips. The problem: I have a 90 minute session where voice talent read words or short phrases in a recording booth. My end goal is for each clip to be in a separate file. There is a regular amount of 'silence' between each take. I do not want regions created for silence. I can imagine how a couple more parameters added to auto region would accomplish this goal. Or a custom script may have already been written that does this. I searched but I couldn't find anything. I just finished a frantic day of marking regions on a 90 minute file and extracting regions. It is brain melting work that a computer should be able to do without intervention. The current auto region is rather 'music-centric' and while I have gotten close I have not found any combination of parameters that will do the job right. Any input would be appreciated. If this is not forthcoming (or done already) then I might be motivated to conquer enought of the scripting language to do it myself as this is a task that I end up doing a lot. thanks in advance ddt
Subject:RE: better auto region for speech Reply by: David Thiel Date:3/23/2010 5:13:46 PM update: While waiting I downloaded the Adobe Trial version of Audition. It has a "Find Phrases and Mark" function that works really well. It cut the phrases nice and tight without truncating anything. Over 30 takes it created two tiny false regions (easily removed with sorting). I have been using Sound Forge loyally since the beginning (1993) and I will continue to use it as it is second nature for me but given the nature of my business, I will have to consider getting a real copy of Audition even if I only use it for speech extraction it will pay for itself in no time. Still I wish that SF & Sony would compete in this area. Game guys are faced with massive speech cut-ups all the time and I would prefer to keep using SF. hiho ddt

Subject:better auto region for speech

Posted by: David Thiel

Date:3/23/2010 4:06:41 PM

The auto region tool isn't enough to automatically create clean regions for extracting speech clips.
The problem:
I have a 90 minute session where voice talent read words or short phrases in a recording booth. My end goal is for each clip to be in a separate file.
There is a regular amount of 'silence' between each take. I do not want regions created for silence.
I can imagine how a couple more parameters added to auto region would accomplish this goal.
Or a custom script may have already been written that does this. I searched but I couldn't find anything.

I just finished a frantic day of marking regions on a 90 minute file and extracting regions. It is brain melting work that a computer should be able to do without intervention.

The current auto region is rather 'music-centric' and while I have gotten close I have not found any combination of parameters that will do the job right.

Any input would be appreciated. If this is not forthcoming (or done already) then I might be motivated to conquer enought of the scripting language to do it myself as this is a task that I end up doing a lot.

thanks in advance

ddt

Subject:RE: better auto region for speech

Reply by: David Thiel

Date:3/23/2010 5:13:46 PM

update:
While waiting I downloaded the Adobe Trial version of Audition. It has a "Find Phrases and Mark" function that works really well. It cut the phrases nice and tight without truncating anything. Over 30 takes it created two tiny false regions (easily removed with sorting).
I have been using Sound Forge loyally since the beginning (1993) and I will continue to use it as it is second nature for me but given the nature of my business, I will have to consider getting a real copy of Audition even if I only use it for speech extraction it will pay for itself in no time.

Still I wish that SF & Sony would compete in this area. Game guys are faced with massive speech cut-ups all the time and I would prefer to keep using SF.

hiho
ddt