dear povey,
now i use the online recognition in the voxforge .when i use the ./run.sh --test-mode live , i have a error:
LIVE DEMO MODE - you can use a microphone and say something
The (bigram) language model used to build the decoding graph was
estimated on an audio book's text. The text in question is
"King Solomon's Mines" (http://www.gutenberg.org/ebooks/2166).
You may want to read some sentences from this book first ...
online-gmm-decode-faster --rt-min=0.5 --rt-max=0.7 --max-active=4000 --beam=12.0 --acoustic-scale=0.0769 online-data/models/tri2b_mmi/model online-data/models/tri2b_mmi/HCLG.fst online-data/models/tri2b_mmi/words.txt 1:2:3:4:5 online-data/models/tri2b_mmi/matrix
PortAudio failed to open the default stream
i do not know what means.when i use the ./run.sh,it is all right and have a good result.
thank you very much.
best wishes,
ben
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
This needs to run on a desktop machine where you can speak into a
microphone, that is the point of that demo. Either you are running
remotely or your desktop is not set up right, but i couldn't help.
Dan
dear povey,
now i use the online recognition in the voxforge .when i use the ./run.sh
--test-mode live , i have a error:
LIVE DEMO MODE - you can use a microphone and say something
The (bigram) language model used to build the decoding graph was
estimated on an audio book's text. The text in question is
"King Solomon's Mines" (http://www.gutenberg.org/ebooks/2166).
You may want to read some sentences from this book first ...
online-gmm-decode-faster --rt-min=0.5 --rt-max=0.7 --max-active=4000
--beam=12.0 --acoustic-scale=0.0769 online-data/models/tri2b_mmi/model
online-data/models/tri2b_mmi/HCLG.fst
online-data/models/tri2b_mmi/words.txt 1:2:3:4:5
online-data/models/tri2b_mmi/matrix
PortAudio failed to open the default stream
i do not know what means.when i use the ./run.sh,it is all right and have
a good result.
thank you very much.
best wishes,
ben
I get the same exact error, on what I thought was a 'perfect' build: latest code, on a state-of-the-art linux version. I'm looking for a workaround where we can build against system-supplied libportaudio rather than downloading and compiling it in tools/portaudio, as this has worked for colleagues trying to get past this same ("microphone mode doesn't work") issue. I just want to understand it, and be able to work around it, also, rather than rely on someone else doing a magical hack (building against libraries from a different linux distribution.)
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Cc'ing Vassil in case he does not get this.
Vassil, I wonder whether it's possible to modify the setup to optionally
use the system-supplied portaudio? Even if you don't have time to do this,
perhaps you could point someone else in the right direction.
I get the same exact error, on what I thought was a 'perfect' build:
latest code, on a state-of-the-art linux version. I'm looking for a
workaround where we can build against system-supplied libportaudio rather
than downloading and compiling it in tools/portaudio, as this has worked
for colleagues trying to get past this same ("microphone mode doesn't
work") issue. I just want to understand it, and be able to work around it,
also, rather than rely on someone else doing a magical hack (building
against libraries from a different linux distribution.)
BTW, everyone- it could even be a problem with the code. I am not an
expert in PortAudio so it's hard for me to say, and I haven't had time to
look at it, but if someone on this discussion would take the time to
understand PortAudio and try to figure out what the code is doing with it
and help us improve it, that would be great.
I am also working on a new online-decoding setup in sandbox/online, and
while this doesn't include audio capture yet, one day it might be nice to
be able to demo that, and whoever learns about PortAudio could help with
that also. It might be, of course, that PortAudio is not the right
solution, and I am open to input on that also. [For now, my plan is to
continue to develop online-decoding more as a library, without considering
audio capture and transmission yet.]
Dan
Cc'ing Vassil in case he does not get this.
Vassil, I wonder whether it's possible to modify the setup to optionally
use the system-supplied portaudio? Even if you don't have time to do this,
perhaps you could point someone else in the right direction.
Dan
On Wed, Apr 16, 2014 at 2:54 PM, Eric Riebling er1k27@users.sf.net wrote:
I get the same exact error, on what I thought was a 'perfect' build:
latest code, on a state-of-the-art linux version. I'm looking for a
workaround where we can build against system-supplied libportaudio rather
than downloading and compiling it in tools/portaudio, as this has worked
for colleagues trying to get past this same ("microphone mode doesn't
work") issue. I just want to understand it, and be able to work around it,
also, rather than rely on someone else doing a magical hack (building
against libraries from a different linux distribution.)
Well, AFAIK Linux is not exactly renown for its glitch-free multimedia support, so I guess a 100% flawless out-of-the-box experience might be hard to achieve...
I think using the system-provided PA may not be possible, without changing the code of the online executables, because as far as I remember we are using some "ring buffer" extensions that are not compiled by default.
To those who have problems I would suggest to try to build the PA in tools/ with ALSA support. IIRC you have to install 'alsa-dev' or similar package first, run our tools/portaudio-install.sh(or similar) script and then make the stuff in online{,bin}.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I also get problems with portaudio, from time to time. My current setup uses PulseAudio exclusively, so ALSA/OSS don't work too well. I am not sure if portaudio supports PulseAudio too well. You can try and configure ALSA (which is a chore) or if you give up, simply record yourself in Audacity, save it as a WAV (16-bit 16kHz) and use online-wav-gmm-decode-faster.
For online demos using a microphone source, you can also give a try with the GStreamer demo (./egs/voxforge/gst_demo). Maybe it would be also worth my while to expand the Java client with audio recording capability, since Java is most likely going to work on any system.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
dear povey,
now i use the online recognition in the voxforge .when i use the ./run.sh --test-mode live , i have a error:
LIVE DEMO MODE - you can use a microphone and say something
The (bigram) language model used to build the decoding graph was
estimated on an audio book's text. The text in question is
"King Solomon's Mines" (http://www.gutenberg.org/ebooks/2166).
You may want to read some sentences from this book first ...
online-gmm-decode-faster --rt-min=0.5 --rt-max=0.7 --max-active=4000 --beam=12.0 --acoustic-scale=0.0769 online-data/models/tri2b_mmi/model online-data/models/tri2b_mmi/HCLG.fst online-data/models/tri2b_mmi/words.txt 1:2:3:4:5 online-data/models/tri2b_mmi/matrix
PortAudio failed to open the default stream
i do not know what means.when i use the ./run.sh,it is all right and have a good result.
thank you very much.
best wishes,
ben
This needs to run on a desktop machine where you can speak into a
microphone, that is the point of that demo. Either you are running
remotely or your desktop is not set up right, but i couldn't help.
Dan
On Thu, Apr 3, 2014 at 7:18 AM, wbgxx333 wbgxx333@users.sf.net wrote:
I get the same exact error, on what I thought was a 'perfect' build: latest code, on a state-of-the-art linux version. I'm looking for a workaround where we can build against system-supplied libportaudio rather than downloading and compiling it in tools/portaudio, as this has worked for colleagues trying to get past this same ("microphone mode doesn't work") issue. I just want to understand it, and be able to work around it, also, rather than rely on someone else doing a magical hack (building against libraries from a different linux distribution.)
Cc'ing Vassil in case he does not get this.
Vassil, I wonder whether it's possible to modify the setup to optionally
use the system-supplied portaudio? Even if you don't have time to do this,
perhaps you could point someone else in the right direction.
Dan
On Wed, Apr 16, 2014 at 2:54 PM, Eric Riebling er1k27@users.sf.net wrote:
BTW, everyone- it could even be a problem with the code. I am not an
expert in PortAudio so it's hard for me to say, and I haven't had time to
look at it, but if someone on this discussion would take the time to
understand PortAudio and try to figure out what the code is doing with it
and help us improve it, that would be great.
I am also working on a new online-decoding setup in sandbox/online, and
while this doesn't include audio capture yet, one day it might be nice to
be able to demo that, and whoever learns about PortAudio could help with
that also. It might be, of course, that PortAudio is not the right
solution, and I am open to input on that also. [For now, my plan is to
continue to develop online-decoding more as a library, without considering
audio capture and transmission yet.]
Dan
On Wed, Apr 16, 2014 at 3:11 PM, Daniel Povey danielpovey@users.sf.netwrote:
Well, AFAIK Linux is not exactly renown for its glitch-free multimedia support, so I guess a 100% flawless out-of-the-box experience might be hard to achieve...
I think using the system-provided PA may not be possible, without changing the code of the online executables, because as far as I remember we are using some "ring buffer" extensions that are not compiled by default.
To those who have problems I would suggest to try to build the PA in tools/ with ALSA support. IIRC you have to install 'alsa-dev' or similar package first, run our tools/portaudio-install.sh(or similar) script and then make the stuff in online{,bin}.
I also get problems with portaudio, from time to time. My current setup uses PulseAudio exclusively, so ALSA/OSS don't work too well. I am not sure if portaudio supports PulseAudio too well. You can try and configure ALSA (which is a chore) or if you give up, simply record yourself in Audacity, save it as a WAV (16-bit 16kHz) and use online-wav-gmm-decode-faster.
For online demos using a microphone source, you can also give a try with the GStreamer demo (./egs/voxforge/gst_demo). Maybe it would be also worth my while to expand the Java client with audio recording capability, since Java is most likely going to work on any system.