a little background: I'm faced with converting 3000 IVR scripts for a new PBX & IVR.
Currently the voice scripts exist in .VOX format--AND..they're not 'written-out'/documented.
I humbly ask if anyone knows of a program, such that, I can dump the .vox file in the program and have it produce a text document. If the .vox format is a problem I could probably convert all of them to .wav or whatever.
Yes there are tons of apps/programs which do speech to text in real time, but I want to be able to "upload/dump" the recording into a program, and obtain text.
Can someone point me in the right direction?
Thank you in advance for any sort of comment/help. SF
The problem is that you are hoping to perform generic natural language processing on low quality audio files. Low quality audio data significantly reduces the reliability of natural language processing software. Upsampling your audio files will not improve their content which means poor results even if you did have access to a natural language engine.
Your best bet is to work with a company that performs hybrid machine/human transcription and pay them for each transcriptions. Alternatively you could consider working with Amazon Mechanical Turk and buy some general purpose human effort to get these transcribed. In both of those cases it is likely that VOX files would not work, you would first need to convert them to WAV or MP3 files to make it easier for the third party to use off the shelf software to listen to the prompts.