Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : General Magic -- Ignore unavailable to you. Want to Upgrade?

To: cool who wrote (2997)	7/29/1998 11:58:00 PM
From: Mark Oliver	Respond to of 10081

Hi guys, I've just spent 2 days talking to everyone I could about Speech Recognition, Text to Speech, Language Translation, Voice Security keys, call center management, microphones, etc., etc. My head is swollen to say the least. Now, here's my take on this: <PulsePoint Communications Signs Licensing Agreement With Lernout & Hauspie Speech Products; PulsePoint Communications Leads with Speech-enabled Enhanced Services Solutions> This is an agreement to use LH's TTS. This is not for Speech Recognition. This announcement is nearly meaningless in itself when related to GM. I have learned that there are many competitive offerings for both SR and TTS. In fact, I can say by my ear LH does not have a great TTS and it is not better than Accuvoice. One could argue that there are better offerings than Accuvoice, but actually I never heard one that is commercial. LH has 2 TTS systems. They have a stripped down version for limited power requirements such as palm PC's. It is very hard on your ears, but ok for limited use. Believe me you do not want to hear a lot of e-mail read with this. Serious headache material. LH has a higher grade system that is again in my opinion less pleasing to Accuvoice. The systems are similar in character. Both can be improved upon by the system integrator. I have heard the explanation of how this art of TTS is accomplished, but I won't try to repeat it. It's not important. What is important is TTS is generally not very pleasing. It works, but.... The panel at the end of the show also agreed that TTS was an area in need of great improvement. New systems may appear in the next year or 2 which will greatly change TTS. At that time, I believe GM will be able to simple add the best product. PS I start out saying how this is not about Speech Recognition. In fact, I think LH would not be the best SR choice for this kind of application anyway. There are 2 basic groups for SR. Speaker Dependent and Speaker Independent. Dependent systems are very complicated and accept thousands of words that are learned over time building a library of utterances from a user. These systems are used for dictation. Independent systems are more specific to a use such as stocks, plane schedules, weather, etc. By limiting the word phrases that the system understands, they can work with first time users. They can achieve about 95% accuracy with first time users. Systems like Portico would be Speaker Independent. One can not say that LH doesn't have plans for Independent systems, but their main focus is dependent. Well know Independent systems are Nuance, Applied Language Tech, and Phillips to name a few. Nuance is currently used in systems for Sears, Charles Schwab and American Airlines. I believe Nuance is the SR used by GM for what it's worth. Regards, Mark