Forums

Speech proceesing of Stereo Vs Mono sound?

Started by Unknown January 25, 2015
When speech files are given and turned out that some are stereo, am I correct to assume that the two channels are identical ... hence it is sufficient to just to process one channel and extract features?

So in the code, i just use an index and read all the indices that are odd and use the correct byte size for memory allocation?

I used Audacity to record a 5sec sound - it turned out to have become stereo.

I am also wondering how common are stereo speech sounds nowadays? If not I want to re-record as mono.  Is FLAC stereo?
On 1/25/2015 6:06 PM, speech2020@gmail.com wrote:
> When speech files are given and turned out that some are stereo, am I correct to assume that the two channels are identical ... hence it is sufficient to just to process one channel and extract features? > > So in the code, i just use an index and read all the indices that are odd and use the correct byte size for memory allocation? > > I used Audacity to record a 5 sec sound - it turned out to have become stereo. > > I am also wondering how common are stereo speech sounds nowadays? If not I want to re-record as mono. Is FLAC stereo? >
All audio formats can be stereo or mono. You ended up with stereo because of your Audacity settings at the time. It is trivially easy to convert from one to the other. Audacity can do it, just ask how in the Audacity forum: http://forum.audacityteam.org/index.php My preferred program (for ease) to convert between stereo and mono is "Audio Cleaner II", which you can get, fully featured, free :-) for 30 days. :-( http://speechpro.com/product/noise_cancellation/soundcleaner Use WAV for maximum amount of information, intelligibility and wide compatibility. It is the standard. > I am also wondering how common are stereo speech sounds nowadays? Very. If you want to verify whether both left and right channels are identical, save them in separate files and use this utility: http://www.libinst.com/Audio%20DiffMaker.htm to find any difference. Note: That program only accepts 16-bit audio and is only compatible with Windows XP. -Ramon
On 1/25/2015 6:06 PM, speech2020@gmail.com wrote:
> I am also wondering how common are stereo speech sounds nowadays?
I have about zero experience with Speech Recognition, but I doubt that there will be much difference in your error/recognition rates. Your stereo file will be twice as large as a mono one. You may get lucky (luckier than myself, anyway) and be able to get help here: http://cmusphinx.sourceforge.net/wiki/communicate/ -RFH
On 1/25/2015 6:06 PM, speech2020@gmail.com wrote:
> When speech files are given and turned out that some are stereo, am I correct to assume that the two channels are identical ... hence it is sufficient to just to process one channel and extract features? > > So in the code, i just use an index and read all the indices that are odd and use the correct byte size for memory allocation? > > I used Audacity to record a 5sec sound - it turned out to have become stereo. >
There you go... https://www.youtube.com/watch?v=ZJ-Nq-os2YI -RFH
How about using the average of left and right channel?	 

_____________________________		
Posted through www.DSPRelated.com
On Sunday, January 25, 2015 at 5:28:08 PM UTC-8, Ramon F Herrera wrote:
> On 1/25/2015 6:06 PM, wrote: > > When speech files are given and turned out that some are stereo, am I correct to assume that the two channels are identical ... hence it is sufficient to just to process one channel and extract features? > > > > So in the code, i just use an index and read all the indices that are odd and use the correct byte size for memory allocation? > > > > I used Audacity to record a 5sec sound - it turned out to have become stereo. > > > > > There you go... > > https://www.youtube.com/watch?v=ZJ-Nq-os2YI > > -RFH
Thanks Ramon. Appreciate it - curious do u have an online profile?