I've never had a system previously which has had effective voice recognition, and hence good speech dictation facilities. The new iPad Pros have Siri, which has been available on iPhones for years. As implemented on the iPad Pro, it does seem to require an internet connection.
Some "explanations" have been given previously, in relation to the iPhones - https://discussions.apple.com/thread/3452329?start=15 Of course, as technology develops, at any one time there are better ways of doing some things than others. A few years ago, it may have been the case that large organisations could install powerful, fast servers, perhaps with specialised hardware and software, to get some things done quickly. This has been a common feature of computer technology for decades, and thus there have been many attempts to do processing remotely, rather than locally - partly based on assumptions that the overall effectiveness of
: send data to remote server - process on remote server - send response back to local device
is better than
: do all processing locally with local data.
For some applications external data is required anyway, so one argument relating to Siri is that data is going to have to be transmitted to remote servers in any case, so the speech to text translation can be done there, in addition to any other processing which might be needed.
The downside of using tools such as Siri is precisely that data is exported, and further that the detailed nature of the data is not known by the users.
It clearly should be possible to do good speech to text translation locally, if the local device has sufficient computing power, storage space, and runs good algorithms, which in 2016 should at least be possible for many consumer computer tools, including some portable devices.
Over the history of electronic computing so far, there have been numerous attempts to shift the balance betweeen local and remote computing. Many have effectively failed, due to the rapid development of more powerful devices which can be bought cheaply and used locally. The thin client models and display only devices in the twentieth century mostly failed because it gradually turned out to be cheaper to make proper computers, rather than only displays, and the separation of functionality between local and remote systems, plus the overheads of relying on network links, just did not work out.
Simple tests show that Siri can be used to write text using word processing packages such as Pages, and in email, and it does appear to be quick. It is also possible to swear and use obscene words, and these may well be rendered correctly. Dr Johnson omitted some words from his dictionary, and is reported to have said on at least one occasion "Madam, you looked for them!" in relation to some "questionable" utterances.
See also http://grammar.about.com/od/words/a/...Dictionary.htm where a simlar anecdote is alleged relating to two ladies.
While this may be amusing, it might not be a good idea to do this frequently, nor to use such voice recognition tools to create documents which may be intended to be completely private.
Some "explanations" have been given previously, in relation to the iPhones - https://discussions.apple.com/thread/3452329?start=15 Of course, as technology develops, at any one time there are better ways of doing some things than others. A few years ago, it may have been the case that large organisations could install powerful, fast servers, perhaps with specialised hardware and software, to get some things done quickly. This has been a common feature of computer technology for decades, and thus there have been many attempts to do processing remotely, rather than locally - partly based on assumptions that the overall effectiveness of
: send data to remote server - process on remote server - send response back to local device
is better than
: do all processing locally with local data.
For some applications external data is required anyway, so one argument relating to Siri is that data is going to have to be transmitted to remote servers in any case, so the speech to text translation can be done there, in addition to any other processing which might be needed.
The downside of using tools such as Siri is precisely that data is exported, and further that the detailed nature of the data is not known by the users.
It clearly should be possible to do good speech to text translation locally, if the local device has sufficient computing power, storage space, and runs good algorithms, which in 2016 should at least be possible for many consumer computer tools, including some portable devices.
Over the history of electronic computing so far, there have been numerous attempts to shift the balance betweeen local and remote computing. Many have effectively failed, due to the rapid development of more powerful devices which can be bought cheaply and used locally. The thin client models and display only devices in the twentieth century mostly failed because it gradually turned out to be cheaper to make proper computers, rather than only displays, and the separation of functionality between local and remote systems, plus the overheads of relying on network links, just did not work out.
Simple tests show that Siri can be used to write text using word processing packages such as Pages, and in email, and it does appear to be quick. It is also possible to swear and use obscene words, and these may well be rendered correctly. Dr Johnson omitted some words from his dictionary, and is reported to have said on at least one occasion "Madam, you looked for them!" in relation to some "questionable" utterances.
See also http://grammar.about.com/od/words/a/...Dictionary.htm where a simlar anecdote is alleged relating to two ladies.
While this may be amusing, it might not be a good idea to do this frequently, nor to use such voice recognition tools to create documents which may be intended to be completely private.
Comment