on December 22, 2013
Right now, this product is showing "unavailable," and I hope it stays that way. I'm only posting this in case it comes back, to save others the hassle. (I purchased straight from Nuance website.) Here's my breakdown:
1. It will not accept MP3 files, despite their being the most common audio format. You need to convert them. Not a super big deal for me, but something some people won't be able to do, and ridiculous since MP3s are so common.
2. Many, many crashes during training process, which leads to lost "training" and appears to make the software even worse at recognizing the speech.
3. Files must be of a certain bit-rate quality, which (as with format conversion) many people may not know how to do or convert to.
4. After literally HOURS of "training" the software, it produced no improved results, with the transcription being 75-90% useless. And let me tell you, the training process is tedious. (It does a test transcription, and you go in a type-correct all of its errors.) Ironically, as I mentioned above, the software was worse after training, though I don't know if this was because of the crashes. However, what difference does that make in the end?
5. My guess, from my experience and reading others', is that you should definitely NOT buy this software if you only have 2-3 hours of audio to transcribe and don't expect any more in the future. The money and time you'll spend on this would be better put toward paying a human transcriptionist (approx. $1 per audio minute).
6. Simply don't expect this thing to do with well natural speech, despite all the maker's promises. I'm guessing that it MIGHT do okay, with all the hours of training, if you were to dictate your own audio files in a slow and steady speech. However, the product's main promise (compared to Dragon and dictation) is that it will transcribe audio files -- implying that you can use it for files you already have!! In my case, all I wanted was to transcribe a few hours of interviews between two people... good, clean, simple audio that's high quality. But this software can't hack it.
7. Tech support sucks. Not available except during business hours, and the online support is nearly useless. The customer forums are mostly people desperately trying to help each other with an actual Nuance employee rarely to be involved. Blind leading the blind, most of the time. Not surprisingly, however, Nuance has SALES support people available via chat and phone 24/7. That should've been a red flag for me.
8. NO TRIAL VERSION. You have to buy and hope for your money back if it doesn't work. Luckily, I got the download version, so they should be able to deactivate it easily enough, to know I'm not continuing to use it.
My ONLY recommendation for buying this software is if you meet the following conditions:
-- You have many hours of existing audio speech you need transcribed.
-- You have the time to put into training the software... hours.
-- The speech in your audio is very clear, crisp, fairly slow, and good English; i.e., it's basically a dictation.
-- Your audio contains one person at a time. Even two people (interview) appears to screw this thing up.
-- You do not expect results that will be any better than 75-80%, despite these best conditions.
-- You are fully prepared to do some serious proofreading and editing, IF you want a polished final result. (This is true, however, for most voice-to-text software.)
And even after all these conditions, you still may not get what you're expecting.