- The Washington Times - Tuesday, October 19, 2010

MacSpeech Scribe is a deceptively good piece of software, well worth the $149.99 you would pay at full retail (though Amazon.com has it for $20 less). If you ever work with recorded speech, this is a program to consider.

Here’s why: MacSpeech Scribe will, with training, take a recording and transcribe, accurately, up to six different “voice profiles” in a given recording. It’ll produce a transcript faster than a human typist, and with stunning accuracy.

And here’s what one might call an “off label” use, if MacSpeech Scribe were a prescription medicine: It doesn’t do half bad when you haven’t trained the software, either. (www.macspeech.com cautions, “transcribed audio must be of a single speaker’s dictation,” but I’ve done tapes with more than one speaker on them).

But let’s start with an orthodox approach: Install the program on any Intel-based Mac, load a recording of your voice (or the one you need transcribed) and play a clip. Go over the transcription (an audio playback feature is included in the program window) and edit the words to match what you’re hearing. That’s it: The program is now “trained.”

The applications for this seem limitless: Transcribing public speeches, class lectures, business meetings and what-have-you. Me? I’ve used it for a few interviews, and it offers great performance and good advantages over doing it all yourself or outsourcing it to another person. It may not be perfect when used to convert the voices of multiple speakers into text, but you’ll have a transcript that is far easier to edit, I believe, than trying to type the whole thing yourself.

And even if you restrict MacSpeech Scribe to its “authorized” use, transcribing the words of a single speaker, it would represent a tremendous advancement in the world of speech-to-text transcribing. Hours and hours would easily be saved, and from what I’ve seen, overall accuracy is stunning.

The program is very much plug-and-play, in the sense that a simple “quick start” sheet will get you installed, trained and running with the software. Help files and online assistance are available, but I’ve found little need for either in the course of using this software.

Where to get the audio files for transcription? You can start, of course, with those micro-sized digital recorders that are available for about $100, but if you have an Apple iPhone, the “voice memo” feature works just fine. (Just the other day, for example, I used my iPhone as a “backup” to a cassette recording. The tape failed — don’t ask me why — but the iPhone worked like a charm.)

I would imagine that audio tracks from recorded lectures or podcasts, if converted to the proper formats — the program supports .wav, .aif, .aiff, .m4v, .mp4, or .m4a audio files — would be equally suitable.

So much of history today is verbal, and so many things can be conducted verbally such as oral histories and memoirs, that having this program will be of invaluable use to many. I can only hope the makers will expand its capabilities to officially transcribing multiple speakers: at that point, this could become breakout software for the nation’s law firms, perhaps even for Congress and its seemingly innumerable hearings.

Although Im still waiting for the telephone-call-to-computer speech transcriber, MacSpeech Scribe is just about the next best thing.

I’m very impressed by MacSpeech Scribe, and its performance will make me want to test some of Nuance Communications’ other software products, such as the Dragon Naturally Speaking line, which specializes in Windows computer control and typing by voice commands. You’ll find my results here, whether or not I type them on a keyboard.

E-mail mkellner@washingtontimes.com.