[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Accents and low-pass filtering



Dear Aniruddh and Etienne, 

There is a Free & Open Source Software command line tool called "youtube-dl", 
which can download a youtube video, extract the audio and convert it to a 
desirable format using ffmpeg. Youtube-dl can be installed in Unix and Windows
platforms. 

In MAC OS enviroment, the installation could be done using macports (which 
also can be used for ffmpeg installation). 

More information at:

Macports: http://www.macports.org

Youtube-dl: http://rg3.github.com/youtube-dl/

Hope that helps.
Kind regards,
Konstantinos Drossos

On 8 ÎÎÏ 2013, at 9:58 Ï.Î., Etienne Gaudrain <e.p.c.gaudrain@xxxxxxx> wrote:

Dear Aniruddh,

Sounds great!

You can download the video using online services or plugins for you browser. Then it should be either in mp4 or flv format.

To extract the audio, you can use command line tool ffmpeg (will also be able to convert to wav in the process).

The same can be used to convert the video into a format that Matlab will be happy with (could be just a collection of images).

Load the sound in Matlab (wavread), and low pass. Load the video in Matlab (either by making a big matrix with the images, or by using the VideoReader class). You can then probably also just low pass. I guess this will blur any fast motion. For that reason you may want to convert to a different color space first (e.g. rgb2hsl).

Produce either a video or a collection of images with Matlab.

Then recombine audio and video using ffmpeg.

You should find the command lines for ffmpeg by Googling around.

Hope that helps,
-Etienne



On 7 March 2013 21:45, Aniruddh Patel <apatel@xxxxxxx> wrote:
Dear Colleagues,

I would like to create a version of this short YouTube video (of an
actress doing 21 different accents) in which the audio is low-passed
filtered, so that students can focus on the prosody of the different
accents:

http://www.youtube.com/watch?v=3UgpfSp2t6k

I would also like to create another version in which both the audio and
the video are low pass filtered, so that one sees the slower time-scale
movements of the face while hearing the low-pass filtered voice.

Can anyone suggest how I might do this?

Regards,

Ani Patel

Aniruddh D. Patel
Associate Professor
Dept. of Psychology
Tufts University
490 Boston Ave.
Medford, MA 02115

a.patel@xxxxxxxxx
http://ase.tufts.edu/psychology/peoplePatel.htm