How Can Someone Accomplish This?

  • Thread starter Thread starter mrplow
  • Start date Start date
M

mrplow

New member
I got this email from a friend, and I figured if anyone knew the solution, it'd be someone on this board. Any suggestions from the Home Recording folks out there?

(email pasted below)
--------------------------------------

I am wondering if you have an answer to a problem I have. I have been trying to think of a way to break up a sound file. Assuming I have a recording of a reading, I would like to break it apart acccording to where the end of the sentence is. I have been thinking that I would have to need sound recognition and another piece of software to break it apart. But I don't think the sound recognition would be very reliable, since it would be hard to differentiate between comas and periods.

Now, since none of these recordings exist yet, I was thinking that if I record the readings with a sound at the end of each sentence, say, a chime. I would have a repetitive pattern within the file (one distinctive sound to tell the software where the sentence ends), it would make it easier to break the file apart. Now, the reason I want to do this is because I want to do it automatically, not manually. Why? Well, becuase I might have more than 8000 of these files and doing it manually would be a pain in the neck. Do you have any suggestions on this kind of software (to break apart a file into bits and automatically save the results)? If you have done something like this, or have a good suggestion, it would be amazing. Thanks.

-------------------------------------

MrPlow
 
I know Reaper has an auto-trim/split function with a variable threshold that would probably work. One would have to check your work anyway, so why not just go ahead and do it manually? It'll take a long time and won't be fun, but you'll know it's done right.
 
Will whoever is doing the reading be adding the chime as well? Or will the chimes be added afterward?
 
Woah, I don't envy the person who's got to edit that.

Personally I'd edit it manually, simply because I don't trust automated procedures for editing voice recordings. Sending 8000 files to a client where say 4000 are wrong doesn't sound too great a situation.

For manual editing I'd use Adobe Audition, which is IMO the best for editing speech into separate files, simply because you can programme the keys, including an ingenious 'Save Selection As' key, which will open a save dialogue for the selected chunk of audio - VERY fast if you got a lot of files to edit. You would be looking at 100-150 files per hour of editing. I've done computer games jobs (1000's of files) in the past with this and it's the best.

Another option, if you don't want to edit manually, is to add markers as you record (I think you can in Audition), then divide it using the markers automatically in Sound Forge, which I've tried once and it works.
 
Record it into waverepair, leaving a second or two where you want the breaks as you go or hitting the button for track marking as you go.
When done go to find tracks. the prog will search for gaps & drop a draft sep into the gap. You zoom in, run through the file looking for the odd error & adjusting as you go. When done (& you can compress, amplify, normalize, EQ, mono, wide stereo etc whilst in there.
Then select split tracks - tell it where to save them to & you have your sentences, phrases, or whatever neatly split, numbered & stored for use as you wish.
Waverepair is about US$30 to download. I use it for LP restortaion but also used it to break up a speech into chucnks so I know it can/will work.
 
Bulls Hit said:
Will whoever is doing the reading be adding the chime as well? Or will the chimes be added afterward?
Click to expand...

Hello!... It would be just as easy to make the break in the sentance as it would be to add in the chime.
 
Agent47 said:
Another option, if you don't want to edit manually, is to add markers as you record (I think you can in Audition), then divide it using the markers automatically in Sound Forge, which I've tried once and it works.
Click to expand...

I like this option best. I would just record straight into Sound Forge, drop markers as you record (you'll probably have to be fast unless you tell the reader to give you a decent pause after each sentence), do Markers to Regions, and then Extract Regions to seperate wav files. Should be fairly painless...
 
Any software that lets you split files or regions via a volume threshold will work. The key is to provide the software with a good cue. What I would do personally is make sure to pause for an extra-long time after each sentence. Say, pause for at least one second after each sentence. Then, set your auto-split function to split files when the audio falls below a certain threshold for more than, say, 800 milliseconds. A normal pause in a sentence is never going to be more than maybe 400ms, so you should be safe this way.

As for the specific details of how to do this, it's different for every DAW. I know how to do it with Pro Tools but I doubt you are using PT.
 
if you can find a WAV Recorder program,
similar to Creative Recorder program (that comes with a Soundblaster) the method would be this :

1) set your input ( LINE IN etc)

2) click on Record button
3) do your sentence
4) Click on Stop
5) Click on SAVE

... repeat 2 to 5 over and over, until completed.
NOTE: .. a new file will be created incrementing with a 1 at the end, then 2, then 3 ....................

Do some research on what program will do this
 
I would just have them mute the input where they want it split and set that as a threshold since the mute will be much quieter than a pause with background noise
 
Thanks for all the suggestions, guys. I'm going to recommend using the Sound Forge + markers option to my friend (and forgo the bloody chimes as well). Cheers.

MrPlow
 
You must log in or register to reply here.
Back
Top