Hi,
This Coursera course (freely available in audit mode) will help:
https://www.coursera.org/learn/audio-signal-processing
cheers
On Thu, Jul 20, 2017 at 9:43 AM, Shashidharan Nair <
ck.shashidharan@xxxxxxxxx> wrote:
My guess is that a little point-to-point algebra can help. This will work
best if there is a point-to-point correspondence between the two files
(identical sampling rates). I am not very optimistic myself but really
curious to know what happens.
k1(original) - k2(background) = k3Vocal. ( of course you should run this
algebra for every single sample)
K1, k2 and k3 are arbitrary constants. K3 is not important but the
other two have to be obtained by a little trial and error.
On 20 July 2017 at 09:32, Atanu Mandal <atanubuieit29@xxxxxxxxx> wrote:
Respected Sir ,
Thanks for your suggestion. Actually I am working on the speech to text
processing problem. I started working in Octave and python. During my work
I found that due to the presence of music or we can say background noise
the speech to text is not giving the expected result. So I decided to keep
only the vocals part. During the process till now what I did is I am able
to remove vocals from the audio. So I have two audio file one is main audio
file and another is only the background sound part.
Till now I am able to do this part only. Hope you got a brief idea of my
work.
On 20-Jul-2017 9:21 AM, "Shashidharan Nair" <ck.shashidharan@xxxxxxxxx>
wrote:
From the physics point of view, in a stereo recording, *generally* the
audio has the same phase in both channels whereas the other sounds are
spatially distributed so each will *generally* have different phases in
the L and R channels respectively.
I have based my arguments on my experience with a hardware solution to
the same problem. Maybe you can try finding the arithmetic difference
between the two channels, scale up the result a little a little and listen
to the tracks. If you find the vocal audio missing then you know you are on
the right track. Let me warn you that what i am telling you is some
educated guess work. Please let me know if my suggestion works and then
maybe we can take this further.
I request other members also to comment on this.
regards
On 19 July 2017 at 19:41, Atanu Mandal <atanubuieit29@xxxxxxxxx> wrote:
Respected all,
I guess this is not the right place to discuss about my problems
however I am sending this mail for at-least some suggestions.
I was trying to work with audio files using python. I need to keep
vocals of an audio file removing the background sounds. I am stuck with the
problem. Can anyone suggest how to proceed?
--
Thanks and Regards,
Atanu Mandal
Contact : +91 9062866035