Difference between revisions of "Talk:LlSpeech2Text"

From Second Life Wiki
Jump to navigation Jump to search
 
(3 intermediate revisions by 2 users not shown)
Line 2: Line 2:
* For the matter, how would it handle different accents in English? Such a function would be pretty much impossible as proposed. To make any sort of sense, the function should be able to accept a list with expected inputs, and the function would then send the closest match. However, it would still require WAY too much CPU time. Also, remember that voice doesn't occur on LL's servers (as far as I know) - it's provided by another company. In short, while I can see where the idea is coming from, I highly doubt it will ever be implemented. --[[User:Kiwi Alfa|Kiwi Alfa]] 12:26, 3 April 2007 (PDT)
* For the matter, how would it handle different accents in English? Such a function would be pretty much impossible as proposed. To make any sort of sense, the function should be able to accept a list with expected inputs, and the function would then send the closest match. However, it would still require WAY too much CPU time. Also, remember that voice doesn't occur on LL's servers (as far as I know) - it's provided by another company. In short, while I can see where the idea is coming from, I highly doubt it will ever be implemented. --[[User:Kiwi Alfa|Kiwi Alfa]] 12:26, 3 April 2007 (PDT)


Why would it have to run on the sim?... It would just act like a listener and receive chat-like input from the client on a special channel like the debug except for voice that was converted to speech. The client wouldn't need to send the voice text, only when told to do so. --[[User:Nexii Malthus|Nexii Malthus]] 16:37, 22 July 2008 (PDT)
Why would it have to run on the sim?... It would just act like a listener and receive chat-like input from the client on a special channel like the debug except for voice speech that was converted to text. The client wouldn't need to send the voice text, only when told to do so. Also how it would handle languages other then english completely depends on the speech2text code. Oh and we will anyways have a function to determine an agent's language quote soon. --[[User:Nexii Malthus|Nexii Malthus]] 16:37, 22 July 2008 (PDT)
*Interface language != language spoken... I could have switched between English, Chinese and Japanese in one session just so the code would have a hard time understanding which language is being spoken. Let's not even get started with languages like Lojban... --[[User:Geneko Nemeth|Geneko Nemeth]] 18:09, 22 July 2008 (PDT)
 
How would it handle misspelled words? -- '''[[User:Strife_Onizuka|Strife]]''' <sup><small>([[User talk:Strife_Onizuka|talk]]|[[Special:Contributions/Strife_Onizuka|contribs]])</small></sup> 01:06, 23 July 2008 (PDT)

Latest revision as of 01:06, 23 July 2008

I think this feature would be very CPU intensive for the sim to run. How would it handle languages other then english? Strife Onizuka 11:40, 3 April 2007 (PDT)

  • For the matter, how would it handle different accents in English? Such a function would be pretty much impossible as proposed. To make any sort of sense, the function should be able to accept a list with expected inputs, and the function would then send the closest match. However, it would still require WAY too much CPU time. Also, remember that voice doesn't occur on LL's servers (as far as I know) - it's provided by another company. In short, while I can see where the idea is coming from, I highly doubt it will ever be implemented. --Kiwi Alfa 12:26, 3 April 2007 (PDT)

Why would it have to run on the sim?... It would just act like a listener and receive chat-like input from the client on a special channel like the debug except for voice speech that was converted to text. The client wouldn't need to send the voice text, only when told to do so. Also how it would handle languages other then english completely depends on the speech2text code. Oh and we will anyways have a function to determine an agent's language quote soon. --Nexii Malthus 16:37, 22 July 2008 (PDT)

  • Interface language != language spoken... I could have switched between English, Chinese and Japanese in one session just so the code would have a hard time understanding which language is being spoken. Let's not even get started with languages like Lojban... --Geneko Nemeth 18:09, 22 July 2008 (PDT)

How would it handle misspelled words? -- Strife (talk|contribs) 01:06, 23 July 2008 (PDT)