[Project-ideas] Project Proposal for GSOC2013

[Sankarshan Mukhopadhyay]

On Wed, Apr 17, 2013 at 5:59 PM, Shikhar Sharma
<sharma.shikhar at yahoo.com> wrote:
> I want to propose a project which combines the use of speech recognition and
> speech synthesis.

I can see how this would be useful. Your example below illustrates one
possibility. However, one of the aspects is that your example
considers to Latin-based languages. Additionally, I have another query
(in-line)

>    Let say if I have a famous French video, but unfortunately I don't
> understand French
>    language. If I am desperate to watch this video, the only option in hand
> is to search
>    for a dubbed version in English, as I can understand English. But what if
> the dubbed
>    movie is not available??
>    So to help this situation out, I am planning to develop a plug-in which
> can be installed in
>    any of the video players say VLC. The plug-in will then convert the audio
> from one language
>    to another. To accomplish this, I will use speech recognition to
> translate
>    the spoken words into text, say from French audio to French text.

I am guessing that there is a stage here where you use a continuous
translation system to translate text in language A into a text in
language B (French -> English)

>    Speech Synthesizer would be further used to convert text to speech (TTS).
> So the French text
>    is now converted to English text and which will further be converted to
> English audio using
>    TTS.

Your proposal would require - [1] more detail, [2] assessment of
existing tools, [3] analysis of the gaps [4] assessment of which
aspects you want to make part of your GSoC proposal. At this stage, it
looks simple enough to tie together a couple of disparate systems, but
I am guessing that this is not so. First looks are a bit deceptive.


--
sankarshan mukhopadhyay
<https://twitter.com/#!/sankarshan>