Hi Donna and Aman, I think it's not that what you're looking for doesn't exist, but that there aren't commercially available solutions. Back in 2005-2006, shortly after the original MacVisionaries list got started, there was a podcast search engine named PodZinger, later renamed EveryZing. I think it must have been running a version of the continuous speech recognition system that the company responsible for this effort, BBN, started developing about a decade earlier. At that time the number of broadcast podcasts was much smaller than now. The PodZinger search engine let you type in a phrase or set of keywords, and then it would pull up a match to identified podcasts, and even estimate the time the phrase occurred within the podcast. It was sort of like doing a Google search for podcast audio content, and pretty impressive. You had to type in enough words in the search term to identify the context, because just like a Google search you'd get a short section of matched content, but you didn't have to really type more than you would for a Google search. I think this service was only around for a couple of years.
Probably this was an outgrowth of Department of Defense funded research. You ca probably do a web search to read more details. I don't know of anything like that exisiting commercially, and you'd probably need to have a huge training set (like the database of Siri users with different accents and speech patterns) to train the software. HTH. Cheers, Esther On Wednesday, July 17, 2013 1:51:34 AM UTC-10, Donna wrote: > Hi, Aman, > > > > Unfortunately, it was the latter. I kind of didn't think that there was > anything that could do this, but I figured if it was out there, someone on > this list would know about it. > > > > thank you for responding, if nothing else, it's good to be sure that what I > was looking for doesn't exist. > > Cheers, > > Donna > > On Jul 16, 2013, at 2:47 PM, Aman Singer wrote: > > > > > Hi, Donna. > > > If I may ask, what sort of speech are you looking to convert? That is, > > > are you looking to convert speech from a speaker over which you have > > > control, or recorded speech from a person who is willing to read > > > training text? Alternatively, are you looking to convert speech that > > > is, for example, broadcast, recorded from a speaker who will not train > > > the software, or some other speaker over which you don't have any > > > control? The first is fairly simple. If you can have the speaker > > > record his/her/its training speech on to a digital recorder, there are > > > programs which you can train using that recorded speech and they will > > > then recognise that particular speaker's recorded voice fairly well. > > > If, however, you're after the second, for example, transcribing a > > > broadcast recording, I know of nothing that will produce an acceptable > > > transcription without human input. If you find such a thing, however, > > > I, along with quite a few other people, would be overjoyed, this, > > > particularly in real-time, would be a godsend to those of us with bad > > > hearing. If you find anything like this, then, please let the list > > > know. > > > Aman > > > > > > > > > On 7/16/13, Donna Goodin wrote: > > >> Hello all, > > >> > > >> Does anyone know of any software that will take speech, not dictation but > > >> recorded speech, and converted to text? It could either be mobile software > > >> or software for the Mac. > > >> Thanks, > > >> Donna > > >> > -- You received this message because you are subscribed to the Google Groups "MacVisionaries" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/macvisionaries. For more options, visit https://groups.google.com/groups/opt_out.
