FLV => FFMPEG => SPHINX4 problem

[Ben Norbury]

Hello

I am just getting to grips with Sphinx 4 and it is looking quite promising for my needs.

This works:
I record from my machine's webcam an audio clip using windows sound recorder at 22Khz. I upload and pass it through FFMPEG to reduce it down to 16Khz. I then run it through Sphinx 4 and it gives me back some words, great!

Doesn't work:
I record my sound using the same webcam but it is now streamed to the server and saved as an FLV file at 22Khz. I do the same, pass it through FFMPEG and reduce it to 16Khz. I now pass it through Sphinx 4 at there are no matches.

The clips look the same when I view the audio details:

Works:
WAVE (.wav) file, byte length: 3260104, data format: PCM_SIGNED 16000.0 Hz, 16 bit, mono, 2 bytes/frame, little-endian, frame length: 1630030

Doesn't Work:
WAVE (.wav) file, byte length: 723744, data format: PCM_SIGNED 16000.0 Hz, 16 bit, mono, 2 bytes/frame, little-endian, frame length: 361850

BTW when I say it doesn't work, the file is accepted but nothing is matched.

I am running 1.0Beta3 on Linux.

Any help would be really appreciated.

Thank you
Ben

---

If it helps, I can post it all if needed...

<property name="absoluteBeamWidth" value="5000"/>
<property name="relativeBeamWidth" value="1E-120"/>
<property name="absoluteWordBeamWidth" value="200"/>
<property name="relativeWordBeamWidth" value="1E-80"/>
<property name="wordInsertionProbability" value="0.2"/>
<property name="languageWeight" value="10.5"/>
<property name="silenceInsertionProbability" value=".1"/>
<property name="frontend" value="epFrontEnd"/>
<property name="recognizer" value="recognizer"/>
<property name="showCreations" value="false"/>

<!-- ******************************************************** -->
<!-- The Dictionary configuration                            -->
<!-- ******************************************************** -->
<component name="dictionary" 
    type="edu.cmu.sphinx.linguist.dictionary.FastDictionary">
    <property name="dictionaryPath"
              value="resource:/edu.cmu.sphinx.model.acoustic.WSJ_8gau_13dCep_16k_40mel_130Hz_6800Hz.Model!/edu/cmu/sphinx/model/acoustic/WSJ_8gau_13dCep_16k_40mel_130Hz_6800Hz/dict/cmudict.0.6d"/>
    <property name="fillerPath" 
          value="resource:/edu.cmu.sphinx.model.acoustic.WSJ_8gau_13dCep_16k_40mel_130Hz_6800Hz.Model!/edu/cmu/sphinx/model/acoustic/WSJ_8gau_13dCep_16k_40mel_130Hz_6800Hz/dict/fillerdict"/>
    <property name="addSilEndingPronunciation" value="false"/>
    <property name="wordReplacement" value="<sil>"/>
    <property name="unitManager" value="unitManager"/>
</component>


<!-- ******************************************************** -->
<!-- The Language Model configuration                         -->
<!-- ******************************************************** -->
<component name="trigramModel" 
    type="edu.cmu.sphinx.linguist.language.ngram.SimpleNGramModel">
    <property name="location" 
        value="resource:/edu.cmu.sphinx.demo.transcriber.Transcriber!/edu/cmu/sphinx/demo/transcriber/transcriber.trigram.lm"/>
    <property name="logMath" value="logMath"/>
    <property name="dictionary" value="dictionary"/>
    <property name="maxDepth" value="3"/>
    <property name="unigramWeight" value=".7"/>
</component>


<!-- ******************************************************** -->
<!-- The acoustic model configuration                         -->
<!-- ******************************************************** -->
<component name="wsj"
           type="edu.cmu.sphinx.model.acoustic.WSJ_8gau_13dCep_16k_40mel_130Hz_6800Hz.Model">
    <property name="loader" value="wsjLoader"/>
    <property name="unitManager" value="unitManager"/>
</component>

<component name="wsjLoader" type="edu.cmu.sphinx.model.acoustic.WSJ_8gau_13dCep_16k_40mel_130Hz_6800Hz.ModelLoader">
    <property name="logMath" value="logMath"/>
    <property name="unitManager" value="unitManager"/>
</component>

[Ben Norbury]

Thanks for the info Nickolay.

[Nickolay V. Shmyrev]

This is not going to work because of audio compression used. It's often ADPCM or CELP at 8 kHz. You need to take a telephone models and adapt them to the audio gone through the channel.

I would better avoid using FLV for transmission/storage.