Programs like Dragon are called voice recognition systems, but clearly that is incorrect. They are not recognizing a persons voice, but rather the spoken word. What is the correct terminology for this type of system?
“Speech recognition” sounds right. “Speech transcription”?
It is voice recognition, though: The software is recongizing a voice and processing it to extract information. It just doesn’t need to be a specific voice.
I don’t see why you feel the need to make a meaningless distinction.
I understand the (albeit pretty pointless) distinction to be that “voice regonition” can be strictly interpreted to be just distinguishing human speech from ambient sound, whereas the Dragon product recognizes the speech to be processed, and then transcribes it into text.
But isn’t there a difference?: voice recognition can determine that my voice is mine, as opposed to yours; as opposed to speech recognition, which recognizes what is being said, not differentiating between who says it.
It’s just analyzing the soundwave in different ways:
Identifying human speech
Differentiating between two or more speakers
Uniquely identifying a given speaker
Analyzing where words/sentences begin/end
Identifying which words were spoken…
Let’s just call it signal analysis?
Because I like to be precise. I differentiate between serpentine and serpiginous as well. Voice recognition is what they use to tell if a voice is that of a particular person. These systems do not recognize voices, they recognize words. Speech recognition sounds about right to me.
Dragon /ScanSoft calls it Speech Recognition.
From Scan Soft web site:
Whether you dictate as a part of your profession, are just learning to type, or want to avoid the pain and tedium of the keyboard, Dragon NaturallySpeaking 8 is the breakthrough in **speech recognition ** that you have been waiting for. A breakthrough in accuracy, delivering an amazing 25% increase and up to an astounding 99% speech to text conversion level. A breakthrough in ease-of-use, allowing you to complete complex tasks with a simple word or phrase. A breakthrough in flexibility in turning recorded speech into text - from Pocket PC’s, Palm Tungsten, Tablet PC and digital recording devices. And finally a breakthrough in affordability, with versions of Dragon NaturallySpeaking tailored to meet your specific needs.
Typically, systems are called “VRUs”…or Voice Response Units. …or voice response software, etc.
Voice Response is what you are looking for.
I think “Voice Response” imply some sort of interaction. I call programs like IBM’s ViaVoice and Naturally Speaking as “speech to text” programs.
When you buy the product, you have to train it to learn your particular voice / speech pattern. There are many variations on how different people pronounce different words, and the program has to learn that sound X tends to mean one thing for person 1 and a different thing for person 2. So in a sense it’s both “voice” and “speech” recognition.
George Orwell called it a speakwrite in the novel Nineteen Eighty Four.
Speech recognition software does voice-to-text translation, determining what words were spoken.
Voice recognition software is used in biometric security applications for identifying the voice of a specific individual.
You train the software for your own voice only to reduce the number of recognition errors. The software cannot tell your voice from anyone elses, nor is it designed to.
Is voice recognition, as defined by TellMeI’mNotCrazy and InvisibleWombat actually used anywhere other than movies? I imagine that there would be much better ways to identify an individual than vocal patterns.
Yes. Do a little Googling (I recommend “biometric” and “voice” as search terms to get started) and you should be able to find some products. It’s real, and you can buy it now.
Their brogue hamster sign two wreck a nice beach.
You confused yourself by mixing two terms, Speech Recognition and Voice Identification. The latter is for security, the first is for computer command and dictation.
No, I didn’t confuse myself at all, Arwin. Granted, I haven’t attended any biometrics conventions lately, but I did quite a bit of research into speech technology in the late 90’s and the term “Voice Recognition” was used extensively to refer to biometric security using voice prints. The term was used by companies in that field at their trade show booths, and in their printed literature.
Lately, a bunch of confused companies and individuals have begun using “voice recognition” and “speech recognition” synonymously. Since I did my work in that field a few years ago, perhaps people in the biometric security biz have coined a new phrase to keep the confusion down.
Fighting ignorance by giving in to it–a noble tradition in the spirit of DSL “modems.” Pah.