Friday, April 27, 2007 6:40 PM
In the previous MSS Beta, I was able to pull out the audio data from the log data in SQL Server and play it back. I took the data from the AudioData table's AudioData field, extracted thimage image data out to a .WAV file and Using the built in ASP.NET SoundPlayer class I was able to play back the audio. It worked great!
I attempted doing exactly the same thing in the latest version of OCS Speech Server Beta and extracted what I think is supposed to be audio data that was imported by the MSSLogToDatabase.exe utility and the data does not appear to be WAV content. In the new Log database schema I am using the EngineRecoAudioData table. The PromptAudioData and NoEngineRecoAudioData tables did not have any data in them. So I concluded the EngineRecoAudioData table was correct.
To test that in fact the ETL files did have audio in them, I used the MSSContentExtract.exe utility to verify that audio was in included in the ETL files, and audio WAV files were extracted and they did play back, but only the user response.
Can you tell me if I need to be doing something different in this version of OCS SpeeechServer Beta that I did not do in the MSS Beta to extract audio from the database? For example are there some switches in the MSSLogToDatabase.exe utility that I need to set to be able to import the audio into the database. Am I using the correct table? Is there maybe some header data I need to append to the Image data I extract from the database to make the audio file I am creating complete?
I also tried using the Analytics and Tuning Studio to listen to the audio, and it too could not play the audio from the database.
Thanks for any information you can provide in this regard.
Friday, April 27, 2007 8:24 PMDid you try createing a new DB and extracting the log data to that?
Saturday, April 28, 2007 1:22 AM
Yes I created a whole new log database and let the schema be automatically created from the MSSLogToDatabase.exe utility.
Thursday, May 03, 2007 9:10 PM
When you import the audio portion using MSSLogToDatabase.exe, you need to add the option /audioession
For more information about the available options: MssLogToDatabase.exe /?
Edit: Should be "/audio: Session" (remove space)
Thursday, May 17, 2007 5:07 PM
Can you let us know the status of your issue? Did you find a solution? If so, would you be able share it with the forums? If not, please let us know your status and if there have been any changes in your environment.
Tuesday, May 22, 2007 1:24 PM
I was able to use the switch you suggested to import the audio (MssLogToDatabase /audioession) into the database and I was able to use the Tuning Studio to listen to the audio. What I have not been able to do is extract the audio using my own code. As before, the audio data that I am obtaining from the EngineRecoAudioData table does not appear to be in the correct format for WAV. Its almost as though the header information is missing or the data is not in a WAV format.
So again I am wondering... Am I using the correct table? Is there maybe some header data I need to append to the Image data I extract from the database to make the audio file I am pulling from the database complete? Is there some possible some sample code that would help?
Thursday, May 24, 2007 12:26 AMFor extracting the user response, one might be able to use EngineRecoAudio.Format (Format chunk) and EngineRecoAudioData.Data (audio data chunk). The lengths and offsets are in the EngineRecoAudioData table if there are many audio data entries for the same speechrequestid.
Thursday, May 24, 2007 1:49 AMThanks for your reply. I will check it out.