Re: speech -> text on FR?
On Mon, Jun 16, 2008 at 6:04 AM, Dan Staley <[EMAIL PROTECTED]> wrote: > I actually just interfaced with the Sphinx project at one of the > research positions I hold. It is actually a very well written interface > (for the most part...there were a few things poorly documented and/or > implemented) But anyway, I found the java version of the project (Sphinx > 4 http://cmusphinx.sourceforge.net/sphinx4/ ) to be pretty easy to > build/interface with. Its great Dan that u got sphinx packages worked for you. I tried it but got some error. However now a days i was concentrating on understanding their some libraries and trying to write my own optimized codes. I will definitely ping you in case of any help. > > > The benefit of using the HMMs and models and methods that Sphinx > implements is that anyone in their programs should be able to specify a > grammar (similar to a simplified regex) that they want to be recognized > and then the interpreter should be able to be user independant...meaning > anyone can speak the phrase into the phone and get the desired output. > Speech training wouldn't be required. I found that once you set it up > correctly, the Sphinx engine is very powerful, and usually identifies > the spoken words no matter who says them (we found it even seemed to > work decently well with a variety different accents). This is good and in fact I will also try to implement this in the model. I will get the HMM models of words by training them from different speakers. This thing i have covered in my Design Document. Thanks in advance... > > -Dan Staley > > On Sun, 2008-06-15 at 19:07 -0400, Ajit Natarajan wrote: > > Hello, > > > > I know nothing about speech recognition, so if the following won't work, > > please let me know (gently :) ). > > > > I understand that there is a project called Sphinx in CMU which attempts > > speech recognition. It seems pretty complex. I couldn't get it to work > > on my Linux desktop. I'm not sure if it would work on an FR since it > > may need a lot of CPU horsepower and memory. > > > > I see a speech project on the OM projects page. To me, it seems like > > the project is attempting command recognition, e.g., voice dialing. > > However, it would be great if the FR can function as a rudimentary > > dictation machine, i.e., allow the user to speak and convert to text. > > > > Perhaps the following may work. > > > > 1. Ask the user to speak some standard words. Record the speech and > > establish the mapping from the words to the corresponding speech. > > It may even be good to maintain separate databases for different > > purposes, e.g., one for UNIX command lines, one for emails, and a > > third for technical documents. > > > > 2. The speech recognizer then functions similar to a keyboard in that it > > converts speech to text which it then enters into the application > > that has focus. > > > > 3. The user must speak word by word. The speech recognizer finds the > > closest match for the speech my checking against the recordings made > > in step 1 (and step 4). The user may need to set the database from > > which the match must be made. > > > > 4. If there is no close match, or if the user is unhappy with the > > selection made in step 3, the user can type in the correct word. A > > new record can be added to the appropriate database. > > > > The process may be frustrating for the user at first, but over time, the > > speech recognition should become better and better. > > > > The separate databases may be needed, for example, because the word > > period should usually translate to the symbol `.' except when writing > > about time periods when it should translate to the word `period'. > > > > I do not know what the storage requirements would be to maintain this > > database. I do not know if the closest match algorithm in step 3 is > > even possible. But if we could get a good dictation engine, that would > > be a killer app, in my opinion. No more typing! No more carpal tunnel > > injuries. No more having to worry about small on screen keyboards that > > challenge finger typing. > > > > Thanks. > > > > Ajit > > > > > > > ___ > Openmoko community mailing list > community@lists.openmoko.org > http://lists.openmoko.org/mailman/listinfo/community > -- Saurabh Gupta Electronics and Communication Engg. NSIT,New Delhi ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: speech -> text on FR?
Hello Ajit, On Mon, Jun 16, 2008 at 4:37 AM, Ajit Natarajan <[EMAIL PROTECTED]> wrote: > Hello, > > I know nothing about speech recognition, so if the following won't work, > please let me know (gently :) ). > > I understand that there is a project called Sphinx in CMU which attempts > speech recognition. It seems pretty complex. I couldn't get it to work > on my Linux desktop. I'm not sure if it would work on an FR since it > may need a lot of CPU horsepower and memory. Indeed the sphinx packages are very well written but they were compiled with the aim of desktop processors. With a lot of data management and large storage, it implements a lot floating point calculations and algorithms. On FR like devices, codes has to be properly adopted and modified. In fact this is the very aim of the GSoC Speech Recognition Project to prepare a speech recognition engine which can run on a processor with 256 or maximum 400MHz without floating point hardware. > > > I see a speech project on the OM projects page. To me, it seems like > the project is attempting command recognition, e.g., voice dialing. > However, it would be great if the FR can function as a rudimentary > dictation machine, i.e., allow the user to speak and convert to text. > yes, once the speech recognition engine is ready then a lot of applications can be built on it. The basic aim of speech recognition will be to identify the work spoken by comparing it with the HMM models of the stored words dictionary and calculating the maximum probability. Once a word has been detected, any API can be called corresponding to that word. > Perhaps the following may work. > > 1. Ask the user to speak some standard words. Record the speech and >establish the mapping from the words to the corresponding speech. >It may even be good to maintain separate databases for different >purposes, e.g., one for UNIX command lines, one for emails, and a >third for technical documents. > > 2. The speech recognizer then functions similar to a keyboard in that it >converts speech to text which it then enters into the application >that has focus. > > 3. The user must speak word by word. The speech recognizer finds the >closest match for the speech my checking against the recordings made >in step 1 (and step 4). The user may need to set the database from >which the match must be made. > > 4. If there is no close match, or if the user is unhappy with the >selection made in step 3, the user can type in the correct word. A >new record can be added to the appropriate database. > > The process may be frustrating for the user at first, but over time, the > speech recognition should become better and better. > > The separate databases may be needed, for example, because the word > period should usually translate to the symbol `.' except when writing > about time periods when it should translate to the word `period'. > > I do not know what the storage requirements would be to maintain this > database. I do not know if the closest match algorithm in step 3 is > even possible. But if we could get a good dictation engine, that would > be a killer app, in my opinion. No more typing! No more carpal tunnel > injuries. No more having to worry about small on screen keyboards that > challenge finger typing. > It would be certainly a great application. But at the moment I am not very sure about the capability of free runner and the applications which it can handle. May be in future more and more betterment can be introduced in the current applications:) > > Thanks. > > Ajit > > ___ > Openmoko community mailing list > community@lists.openmoko.org > http://lists.openmoko.org/mailman/listinfo/community > -- Saurabh Gupta Electronics and Communication Engg. NSIT,New Delhi http://saurabh1403.wordpress.com ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: speech -> text on FR?
On Lun 16 juin 2008 6:00, Brandon Kruse a écrit : > > They also have a sphinx mobile-type of library, which seems to be very > lightweight, and might be worth looking into. > This benchmark (August 2007) compares PocketSphinx, Sphinx 2, 3 (on AMD Athlon 1670 MHz, 512MB of RAM). http://raphaelnunes.wordpress.com/2007/08/08/benchmark-of-sphinx2-sphinx3-pocketsphinx/ Best regards, Gilles ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: speech -> text on FR?
On 16 Jun 2008, at 01:34, Dan Staley wrote: > I actually just interfaced with the Sphinx project at one of the > research positions I hold. It is actually a very well written > interface > (for the most part...there were a few things poorly documented and/or > implemented) Apparently the Openmoko GSoC contributor has also found this: http://lists.openmoko.org/pipermail/community/2008-June/018752.html He's following the list, so I'm sure he'll be along shortly. Hopefully you'll be able to give him some pointers. Stroller. ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: speech -> text on FR?
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Dan Staley wrote: | I actually just interfaced with the Sphinx project at one of the | research positions I hold. It is actually a very well written interface | (for the most part...there were a few things poorly documented and/or | implemented) But anyway, I found the java version of the project (Sphinx | 4 http://cmusphinx.sourceforge.net/sphinx4/ ) to be pretty easy to | build/interface with. | | The benefit of using the HMMs and models and methods that Sphinx | implements is that anyone in their programs should be able to specify a | grammar (similar to a simplified regex) that they want to be recognized | and then the interpreter should be able to be user independant...meaning | anyone can speak the phrase into the phone and get the desired output. | Speech training wouldn't be required. I found that once you set it up | correctly, the Sphinx engine is very powerful, and usually identifies | the spoken words no matter who says them (we found it even seemed to | work decently well with a variety different accents). | | -Dan Staley | | On Sun, 2008-06-15 at 19:07 -0400, Ajit Natarajan wrote: |> Hello, |> |> I know nothing about speech recognition, so if the following won't work, |> please let me know (gently :) ). |> |> I understand that there is a project called Sphinx in CMU which attempts |> speech recognition. It seems pretty complex. I couldn't get it to work |> on my Linux desktop. I'm not sure if it would work on an FR since it |> may need a lot of CPU horsepower and memory. |> |> I see a speech project on the OM projects page. To me, it seems like |> the project is attempting command recognition, e.g., voice dialing. |> However, it would be great if the FR can function as a rudimentary |> dictation machine, i.e., allow the user to speak and convert to text. |> |> Perhaps the following may work. |> |> 1. Ask the user to speak some standard words. Record the speech and |> establish the mapping from the words to the corresponding speech. |> It may even be good to maintain separate databases for different |> purposes, e.g., one for UNIX command lines, one for emails, and a |> third for technical documents. |> |> 2. The speech recognizer then functions similar to a keyboard in that it |> converts speech to text which it then enters into the application |> that has focus. |> |> 3. The user must speak word by word. The speech recognizer finds the |> closest match for the speech my checking against the recordings made |> in step 1 (and step 4). The user may need to set the database from |> which the match must be made. |> |> 4. If there is no close match, or if the user is unhappy with the |> selection made in step 3, the user can type in the correct word. A |> new record can be added to the appropriate database. |> |> The process may be frustrating for the user at first, but over time, the |> speech recognition should become better and better. |> |> The separate databases may be needed, for example, because the word |> period should usually translate to the symbol `.' except when writing |> about time periods when it should translate to the word `period'. |> |> I do not know what the storage requirements would be to maintain this |> database. I do not know if the closest match algorithm in step 3 is |> even possible. But if we could get a good dictation engine, that would |> be a killer app, in my opinion. No more typing! No more carpal tunnel |> injuries. No more having to worry about small on screen keyboards that |> challenge finger typing. |> |> Thanks. |> |> Ajit |> |> | | | ___ | Openmoko community mailing list | community@lists.openmoko.org | http://lists.openmoko.org/mailman/listinfo/community Along with other speex to text engines (as someone else already mentioned), it works best when the engine knows that he could have said something in this list of pre-defined commands, and not any word in general. It is also very good for deciding between two words, eg "yes" or "no", which is more useful than you would think, if you design your interface to the user in the right way. They also have a sphinx mobile-type of library, which seems to be very lightweight, and might be worth looking into. One thing I thought of is when someone tells you a number over the phone, the phone could record and add it to the address book. Lots of cool stuff you could do :) - -brandon -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFIVeVaWSn2Kv7ZyAoRAoOKAJ9V2psUqf9TniZYMUbPp83hvm9lOgCfSaDI qlZ6A+HqDGzZDKpUDaj+oDA= =Wgj4 -END PGP SIGNATURE- ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: speech -> text on FR?
su, 2008-06-15 kello 16:07 -0700, Ajit Natarajan kirjoitti: > I see a speech project on the OM projects page. To me, it seems like > the project is attempting command recognition, e.g., voice dialing. Feasible especially if the user trains the command words in advance. (Didn't check if it does that; it is doable to some extent otherwise too but the difficulty rises markedly...) > However, it would be great if the FR can function as a rudimentary > dictation machine, i.e., allow the user to speak and convert to text. A pipe dream. Save your dictations as audio and postprocess them elsewhere. -- Mikko Rauhala - [EMAIL PROTECTED] - http://www.iki.fi/mjr/> Transhumanist - WTA member - http://www.transhumanism.org/> Singularitarian - SIAI supporter - http://www.singinst.org/> ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
Re: speech -> text on FR?
I actually just interfaced with the Sphinx project at one of the research positions I hold. It is actually a very well written interface (for the most part...there were a few things poorly documented and/or implemented) But anyway, I found the java version of the project (Sphinx 4 http://cmusphinx.sourceforge.net/sphinx4/ ) to be pretty easy to build/interface with. The benefit of using the HMMs and models and methods that Sphinx implements is that anyone in their programs should be able to specify a grammar (similar to a simplified regex) that they want to be recognized and then the interpreter should be able to be user independant...meaning anyone can speak the phrase into the phone and get the desired output. Speech training wouldn't be required. I found that once you set it up correctly, the Sphinx engine is very powerful, and usually identifies the spoken words no matter who says them (we found it even seemed to work decently well with a variety different accents). -Dan Staley On Sun, 2008-06-15 at 19:07 -0400, Ajit Natarajan wrote: > Hello, > > I know nothing about speech recognition, so if the following won't work, > please let me know (gently :) ). > > I understand that there is a project called Sphinx in CMU which attempts > speech recognition. It seems pretty complex. I couldn't get it to work > on my Linux desktop. I'm not sure if it would work on an FR since it > may need a lot of CPU horsepower and memory. > > I see a speech project on the OM projects page. To me, it seems like > the project is attempting command recognition, e.g., voice dialing. > However, it would be great if the FR can function as a rudimentary > dictation machine, i.e., allow the user to speak and convert to text. > > Perhaps the following may work. > > 1. Ask the user to speak some standard words. Record the speech and > establish the mapping from the words to the corresponding speech. > It may even be good to maintain separate databases for different > purposes, e.g., one for UNIX command lines, one for emails, and a > third for technical documents. > > 2. The speech recognizer then functions similar to a keyboard in that it > converts speech to text which it then enters into the application > that has focus. > > 3. The user must speak word by word. The speech recognizer finds the > closest match for the speech my checking against the recordings made > in step 1 (and step 4). The user may need to set the database from > which the match must be made. > > 4. If there is no close match, or if the user is unhappy with the > selection made in step 3, the user can type in the correct word. A > new record can be added to the appropriate database. > > The process may be frustrating for the user at first, but over time, the > speech recognition should become better and better. > > The separate databases may be needed, for example, because the word > period should usually translate to the symbol `.' except when writing > about time periods when it should translate to the word `period'. > > I do not know what the storage requirements would be to maintain this > database. I do not know if the closest match algorithm in step 3 is > even possible. But if we could get a good dictation engine, that would > be a killer app, in my opinion. No more typing! No more carpal tunnel > injuries. No more having to worry about small on screen keyboards that > challenge finger typing. > > Thanks. > > Ajit > > ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community
speech -> text on FR?
Hello, I know nothing about speech recognition, so if the following won't work, please let me know (gently :) ). I understand that there is a project called Sphinx in CMU which attempts speech recognition. It seems pretty complex. I couldn't get it to work on my Linux desktop. I'm not sure if it would work on an FR since it may need a lot of CPU horsepower and memory. I see a speech project on the OM projects page. To me, it seems like the project is attempting command recognition, e.g., voice dialing. However, it would be great if the FR can function as a rudimentary dictation machine, i.e., allow the user to speak and convert to text. Perhaps the following may work. 1. Ask the user to speak some standard words. Record the speech and establish the mapping from the words to the corresponding speech. It may even be good to maintain separate databases for different purposes, e.g., one for UNIX command lines, one for emails, and a third for technical documents. 2. The speech recognizer then functions similar to a keyboard in that it converts speech to text which it then enters into the application that has focus. 3. The user must speak word by word. The speech recognizer finds the closest match for the speech my checking against the recordings made in step 1 (and step 4). The user may need to set the database from which the match must be made. 4. If there is no close match, or if the user is unhappy with the selection made in step 3, the user can type in the correct word. A new record can be added to the appropriate database. The process may be frustrating for the user at first, but over time, the speech recognition should become better and better. The separate databases may be needed, for example, because the word period should usually translate to the symbol `.' except when writing about time periods when it should translate to the word `period'. I do not know what the storage requirements would be to maintain this database. I do not know if the closest match algorithm in step 3 is even possible. But if we could get a good dictation engine, that would be a killer app, in my opinion. No more typing! No more carpal tunnel injuries. No more having to worry about small on screen keyboards that challenge finger typing. Thanks. Ajit ___ Openmoko community mailing list community@lists.openmoko.org http://lists.openmoko.org/mailman/listinfo/community