Speech Recognition Engines

Nested
Re: Can HTK only process the audio files generated by HSLab (the .sig format)?
User: NancyZxll
Date: 3/29/2012 2:17 pm
Views: 159
Rating: 9

Hello Kmaclean,

    The tutorial said it should keep 48000Hz when recording. Can I use 8kHz instead?

    That's the only difference between tutorial's and my setting.

Must I keep 48000Hz for recroding? Why?

--- (Edited on 3/29/2012 2:17 pm [GMT-0500] by Visitor) ---

Re: Can HTK only process the audio files generated by HSLab (the .sig format)?
User: zoezheng
Date: 4/2/2012 5:58 am
Views: 128
Rating: 7

HELLO Nancy:

I'm trying to process music files using HTK. And I have the same problems as yours.

D:\htk>HCompV -C configz -f 0.01 -m -S flist.scp -M data/hmm0 proto
ERROR [+6251] Input file is not in RIFF format
ERROR [+6213] OpenWaveInput: Get[format]HeaderInfo failed
ERROR [+6313] OpenAsChannel: OpenWaveInput failed
ERROR [+6316] OpenBuffer: OpenAsChannel failed
ERROR [+2050] LoadFile: Config parameters invalid
FATAL ERROR - Terminating program HCompV

I have a database consisting 500 songs of mp3 format. And I convert them to wav. When I run HCopy , no error happened.

I've been working on this problem the whole day and haven't work it out yet!

--- (Edited on 4/2/2012 5:58 am [GMT-0500] by ) ---

Re: Can HTK only process the audio files generated by HSLab (the .sig format)?
User: NancyZxll
Date: 4/2/2012 9:35 am
Views: 110
Rating: 8

Hello Zoe,

    I haven't solved this problem yet. Kmaclean suggested me to use VoxForge tutorial to record my data again and uploaded it.

    I am not sure whether it works or not. And I haven't found the essential different recording step(s) between tutorial's and mine.

    I have no idea at all:( I hope to solve this problem as well. It's so wired. I hope anyone can help us....

   Sorry about that for both of us:(

--- (Edited on 4/2/2012 9:35 am [GMT-0500] by Visitor) ---

Re: Can HTK only process the audio files generated by HSLab (the .sig format)?
User: kmaclean
Date: 4/3/2012 4:17 pm
Views: 101
Rating: 9

>    The tutorial said it should keep 48000Hz when recording. Can I use 8kHz instead?

Yes, you can use 8kHz;

or you can record at 48kHz and downsample using a utility like sox

Note: the audio used for AM training must all be the same sampling rate.

--- (Edited on 4/3/2012 5:17 pm [GMT-0400] by kmaclean) ---

--- (Edited on 4/3/2012 5:20 pm [GMT-0400] by kmaclean) ---

Re: Can HTK only process the audio files generated by HSLab (the .sig format)?
User: NancyZxll
Date: 4/4/2012 5:12 pm
Views: 99
Rating: 9

Hello Zoe,

    Have you solved the problem?

--- (Edited on 4/4/2012 5:12 pm [GMT-0500] by Visitor) ---

Re: Can HTK only process the audio files generated by HSLab (the .sig format)?
User: NancyZxll
Date: 4/4/2012 5:16 pm
Views: 129
Rating: 9

Thanks kmaclean.  Do you know what the end-point detection algorithm in HTK? I have used dual-threshold detection, but the result is not good...

--- (Edited on 4/4/2012 5:16 pm [GMT-0500] by NancyZxll) ---

Re: Can HTK only process the audio files generated by HSLab (the .sig format)?
User: zoezheng
Date: 4/4/2012 9:30 pm
Views: 93
Rating: 9

No,I haven't。 €‚My friend give me some new recorded WAV music. The error remains the same.

--- (Edited on 4/4/2012 9:30 pm [GMT-0500] by zoezheng) ---

--- (Edited on 4/4/2012 9:42 pm [GMT-0500] by zoezheng) ---

Re: Can HTK only process the audio files generated by HSLab (the .sig format)?
User: NancyZxll
Date: 4/5/2012 9:40 am
Views: 510
Rating: 9

I know it. I changed 3 record equipments, including Audacity, Matlab, and Windows Recorder, the error still exists. I doubt that HTK cannot process wav file, but only sig format.

If you are very hurry, I think you need use HSLab to record your data again.

Anyway, could you please let me know if you solved the problem?

Many thanks!!

--- (Edited on 4/5/2012 9:40 am [GMT-0500] by Visitor) ---

Re: Can HTK only process the audio files generated by HSLab (the .sig format)?
User: Vovo
Date: 11/29/2012 7:25 am
Views: 1479
Rating: 8

Hi, I had the same problem running Hlab tool. The solution was running Hlab with -F WAV option. Like this:

Hlab -F WAV chair01.wav

 

--- (Edited on 11/29/2012 7:25 am [GMT-0600] by Visitor) ---

Re: Can HTK only process the audio files generated by HSLab (the .sig format)?
User: Fashionee
Date: 9/15/2017 4:10 am
Views: 2054
Rating: 0

Sorry to bother you mate, but it seems that you know a lot about audio format right? I also have a audio playback issue here and may i have your answer? I downloaded a DSS file on the internet (for own job) but it could not be played on my VLC player and I do not know why. I downloaded this audio converter as you can see (http://www.videoconverterfactory.com/tips/convert-dss-to-mp3.html) to convert this DSS audio file. However, before conversion, it asked me to preset parameters, i know i should choose MP3. But how about the bit rate and sample rate? What do these 2 parameters mean? What should i do, does any one know?

--- (Edited on 9/15/2017 4:11 am [GMT-0500] by Fashionee) ---

PreviousNext