VoxForge
VoxForge
I'm testing my acoustic modes uder Linux and using the wav files used for training.But julius seems can not read speech data from most of these files(it decodes 20 out of 100 utterances,but the other 80 are failed).The error infos are shown as follows:
### read waveform input
Error: gzfile: unable to open wav/1.wav
Error: adin_file: failed to open wav/1.wav
Error: adin_file: failed to read speech data: "wav/1.wav "
......
the wav files are recorded under Windows by Audacity.Can't the files recorded under Windows be decoded correctly under Linux?
Does anybody know the reason?thanks
--- (Edited on 3/4/2008 1:06 am [GMT-0600] by kay) ---
Hi kay
>Can't the files recorded under Windows be decoded correctly under Linux?
There shouldn't be a problem. Are you sure you exported your audio to wav and not raw? Does Audacity on Linux recognize the audio files as wav files (wav has a header at the beginning of each file).
>I'm testing my acoustic modes uder Linux and using the wav files used for training.
Just so I am clear, are you trying to recognize speech on a Linux computer from wav files recorded using Audacity in Windows.
Note that the sampling rate and bits per sample of the speech audio files you are trying to recognize (recorded using Audacity on Windows) must match the sampling rate and bits per sample of the audio used to create/train the acoustic models.
Your Julius config file also has settings that must match the sampling rate and bits per sample of the audio used to create the acoustic models.
Hope that helps,
Ken
--- (Edited on 3/4/2008 1:38 pm [GMT-0500] by kmaclean) ---
Hey Kay,
Please check out the properties of your "wav/1.wav" file. I suppose it is a 'stereo type' convert it to 'mono'. It ll work for sure.
--- (Edited on 3/30/2009 2:18 am [GMT-0500] by vishu) ---