One try only?

[demigraff]

[EMG]

I don't know if anyone else is going to say it, but I appreciate all the work you have done. You've been here a long time and are one of the better contributors. Hopefully you can find a volunteer or three to transcribe your work.

[demigraff]

Thanks. I'm sorry that post turned into a mini emotional breakdown; I'm not coping too well with stress at present, and every time I fail to do something it seems like such a big deal.

At least listening to some simple files helped me to calm down; and I've managed to do 2 of my promises. I started with 10 challenges for the new year, most of the form "do (whatever) every day," and so far my best has been 8/10. But today I said I was going to do this file first, because I might not get a chance later, so I ended up with nothing done on any of my challenges before dinner time. I've managed to do [url=http://writing.mrangel.co.uk]some work on my novel[/url] for the 100k-Words challenge, and finished reading a book for the book-a-week challenge, so I don't feel so bad about it.

Thankyou ^,;,^ It really does help to know I get appreciation

And hopefully I'll find some way to give those files a second shot (and figure out why the most recent has so much noise on it).

[EMG]

One suggestion if you aren't already doing it. Try using a USB based microphone. The analog connections on your computer can pick up noise from the computer and corrupt a file even though you are in a perfectly quiet room.

[Snarficus]

I accidently broke protocol and PM'd you without asking permission first. Basically, there maybe a way to actually "automagically" transcribe files. I will make the attempt to (using software) set up a "system" that could work for you. I will get back to you if I can get it to work. (This should work for most "simple" files, but for the more complex (multichannel and overlapped/multilayered files may take special procedures.)

[EMG]

I hope it works, I once tried to record a file while using speech recognition software. The results were so ugly I gave up after a couple tries.

[demigraff]

I've tried a few times with Sphinx (v3, before they went all Java), but never managed to reach any quality that could be even remotely useful. Would be interested to see what a modern system can do. Most of the files I've posted would benefit from re-recording or alternate versions, if I could just get started, so a working recogniser would be a big help :)

[ParanoidLord]

The last time I heard anything about it, speech recognition hit a brick wall at about 80% accuracy, and with the current algorithms used, no amount of brute force could improve it. Now, this was in late 2010 or so, so things may have improved since then, but I'm not sure. Things like Siri on the iPhone suggest things have improved a bit, but things like Youtube's automatically generated captions... don't.

[lazzyz]

There are probably people, like myself, willing to help you try and create scripts from old files. It'd likely be slower than a computer program but still might work!