I'm running Linux, and my sound card is SoundBlaster PCI 128. When I run "sphinx2-demo",
it says initializing and returns to the command prompt after a few moments. When I tried running "sphinx2-simple" instead, it initializes some stuff, but then gives the message:
ERROR: "ad_i386_linux.c", line 144: Audio ioctl(SPEED): 16036, expected: 16000
FATAL_ERROR: "tty-continuous.c", line 224: ad_open_sps failed
Any help would be appreciated.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Some audio systems can't actually _do_ 16kHz exactly, and they return a slightly different value than what you asked for when you set the speed on the device.
That difference should actually make no real difference. In the code you could actually change 16000 to 16036 and it would compile! ... but i suppose we'll need a 'frequency tolerance' to check if the actual rate is close enough to the desired one.
For now, I'd suggest either commenting out that test or setting the default sampling rate (in the code) to 16036.
kevin
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I ran into the same sort of problem (15908Hz) on my system. In trying to work around it I noticed
that "16000" is a MAGIC_NUMBER in a lot of places,
hardcoded in. This would be the first thing to fix. I'll get 0.2 today and try again.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2000-02-05
When I run "sphinx2-simple" similar things happen, but with a different error message:
ERROR: "ad_i386_linux.c", line 306: Audio read error
FATAL_ERROR: "tty-continuous.c", line 133: cont_ad_calib failed
I'm fairly intrigued by this whole speech recognition thing, so as always, any help is appreciated.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I ran into the same sort of problem (48000Hz) on my system. I tried to change the value of DEFAULT_SAMPLES_PER_SEC to 48000. sphinx2-simple didn't work because I haven't change other 16000's to 48000's. sphinx2-server was able to reach the READY mode. However, when I started sphinx2-client and did some tests, sphinx2-server didn't respond. I'm using sphinx2-0.4 on Linux RedHat 7.2.
I've read the documentation in sphinx2.html and it says that -samp only takes in 8000 or 16000. Does this mean that I have to change my audio card? Is there any way to make it work without changing the sound card? I'd appreciate any help.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I'm running Linux, and my sound card is SoundBlaster PCI 128. When I run "sphinx2-demo",
it says initializing and returns to the command prompt after a few moments. When I tried running "sphinx2-simple" instead, it initializes some stuff, but then gives the message:
ERROR: "ad_i386_linux.c", line 144: Audio ioctl(SPEED): 16036, expected: 16000
FATAL_ERROR: "tty-continuous.c", line 224: ad_open_sps failed
Any help would be appreciated.
Some audio systems can't actually _do_ 16kHz exactly, and they return a slightly different value than what you asked for when you set the speed on the device.
That difference should actually make no real difference. In the code you could actually change 16000 to 16036 and it would compile! ... but i suppose we'll need a 'frequency tolerance' to check if the actual rate is close enough to the desired one.
For now, I'd suggest either commenting out that test or setting the default sampling rate (in the code) to 16036.
kevin
I ran into the same sort of problem (15908Hz) on my system. In trying to work around it I noticed
that "16000" is a MAGIC_NUMBER in a lot of places,
hardcoded in. This would be the first thing to fix. I'll get 0.2 today and try again.
When I run "sphinx2-simple" similar things happen, but with a different error message:
ERROR: "ad_i386_linux.c", line 306: Audio read error
FATAL_ERROR: "tty-continuous.c", line 133: cont_ad_calib failed
I'm fairly intrigued by this whole speech recognition thing, so as always, any help is appreciated.
It's failing in the ioctl() call during calibration (getting the volumes and such). Can you record and play audio back using /dev/dsp (not sphinx2)?
kevin
I ran into the same sort of problem (48000Hz) on my system. I tried to change the value of DEFAULT_SAMPLES_PER_SEC to 48000. sphinx2-simple didn't work because I haven't change other 16000's to 48000's. sphinx2-server was able to reach the READY mode. However, when I started sphinx2-client and did some tests, sphinx2-server didn't respond. I'm using sphinx2-0.4 on Linux RedHat 7.2.
I've read the documentation in sphinx2.html and it says that -samp only takes in 8000 or 16000. Does this mean that I have to change my audio card? Is there any way to make it work without changing the sound card? I'd appreciate any help.