Path: csiph.com!usenet.pasdenom.info!aioe.org!.POSTED!not-for-mail From: GGolf Newsgroups: comp.lang.java.programmer Subject: Re: Reading sound output in Java Date: Sat, 26 Oct 2013 15:40:38 +0530 Organization: Aioe.org NNTP Server Lines: 43 Message-ID: References: <8146fa5b-403e-47a3-bc4f-cb76a9a2239d@googlegroups.com> <26c9e48e-7ab1-45be-8869-56e2f6b9f9ca@googlegroups.com> <273e95fc-a887-410f-a84e-15be809da895@googlegroups.com> NNTP-Posting-Host: FVGDoGrfP8ZlHbKxjf17QA.user.speranza.aioe.org Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Complaints-To: abuse@aioe.org User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:17.0) Gecko/20130107 Thunderbird/17.0.2 X-Notice: Filtered by postfilter v. 0.8.2 Xref: csiph.com comp.lang.java.programmer:26432 On 10/26/2013 3:02 PM, Andreas Leitgeb wrote: > GGolf wrote: >> Let me tell you the use case: I want to >> (i) capture a VOIP conversation (which could be between multiple >> participants, or just two participants), and >> (ii) convert that to text using a speech recognizer (using one of the >> various implementations of JSAPI), and >> (iii) display that text on screen. > > Is this going to happen real-time? (I have no idea if consumer-grade > speech-to-text is fast enough for that.) I've seen a couple of demos where this has been done, though I guess the participants have to speak slower than usual for the transcription to make some sense. > > If so, I'd use a second PC (thus with independent sound system) with > a microphone & line-in(connected to a "Y"-split of the headphones' cable) > to capture both outgoing and incoming speech. > Unless you've got some very special sound hardware on the primary (VoIP > running) machine, capturing both input and output (where input isn't > itself mixed into the output) may just not be possible. Hmm...I think you got a point here. I remember that the Stereo Mix option on Windows wouldn't record the input to the microphone unless you would /play it back/ to the speakers. > > PS: note, that recording of conversations may have legal issues, so make > sure that at the start of the recording everyone involved knows it and > agrees to it. (this may not be relevant if the signal is used *only* for > speech-to-text and displaying of text, and neither text nor signal are > saved, but I'm not a lawyer.) Only the text is going to be displayed (the user can save the transcript if they want, but it won't go anywhere by itself). > If you were from NSA, then you probably > wouldn't need to ask here in the first place... > True :)