voice signature characteristics

[Adam Chapman]

Hi,

Im normally an image processing geek but was wondering about voice
recognition systems. Just out of interest really.

What kind of characteristics do you look at when identifying a certain
persons voice? Obviously people with a deeper voice will produce
sounds at a lower frequency than others, but surely you need more than
that.

Adam

[Adam Chapman]

On Nov 1, 9:49 pm, Adam Chapman

Post by Adam Chapman
Hi,
Im normally an image processing geek but was wondering about voice
recognition systems. Just out of interest really.
What kind of characteristics do you look at when identifying a certain
persons voice? Obviously people with a deeper voice will produce
sounds at a lower frequency than others, but surely you need more than
that.
Adam

I started thinking that even though different instruments, e.g. a
trumpet and piano play the same note, you can tell the difference
between the same note from each instrument. It's pretty late in
england right now, so I'll look it up in the morning. What I'm really
interested in is the "features" that one would look for in a
situation. It's obviously very different to image processing problems
because you can't really 'see' the data you are working with to really
construct a meaningful image in your head of the descriptive data
you're working with.

[a***@home.]

Post by Adam Chapman
Hi,
Im normally an image processing geek but was wondering about voice
recognition systems. Just out of interest really.
What kind of characteristics do you look at when identifying a certain
persons voice? Obviously people with a deeper voice will produce
sounds at a lower frequency than others, but surely you need more than
that.
Adam

Sometimes people need help in formulating the question that they would like
to ask. Are you asking about speech/word recognition or are you asking
about actually identifying a specific individual on the basis of their
voice/speech characteristics?

[Adam Chapman]

Post by a***@home.

Post by Adam Chapman
Hi,
Im normally an image processing geek but was wondering about voice
recognition systems. Just out of interest really.
What kind of characteristics do you look at when identifying a certain
persons voice? Obviously people with a deeper voice will produce
sounds at a lower frequency than others, but surely you need more than
that.
Adam

Sometimes people need help in formulating the question that they would like
to ask.  Are you asking about speech/word recognition or are you asking
about actually identifying a specific individual on the basis of their
voice/speech characteristics?

Good point. I was looking at identifying a particular individual,
regardless of what words they are saying, based on their voice
characteristics. I have found some papers on musical instrument
recognition as a start, but would realy like to know what acoustic
features/characteristics are worth looking at

[answerman]

Post by Adam Chapman

innews:f803c82

Post by Adam Chapman
Hi,
Im normally an image processing geek but was wondering about voice
recognition systems. Just out of interest really.
What kind of characteristics do you look at when identifying a
certain persons voice? Obviously people with a deeper voice will
produce sounds at a lower frequency than others, but surely you
need more than that.
Adam

Sometimes people need help in formulating the question that they
would li

ke

to ask.  Are you asking about speech/word recognition or are you
asking about actually identifying a specific individual on the basis
of their voice/speech characteristics?

Good point. I was looking at identifying a particular individual,
regardless of what words they are saying, based on their voice
characteristics. I have found some papers on musical instrument
recognition as a start, but would realy like to know what acoustic
features/characteristics are worth looking at

In my opinion, this is a voice identification issue, not an acoustics
issue. A Google search on "voice identification" should be more than
enough to get you started on the right track.

[Angelo Campanella]

Post by Adam Chapman
What kind of characteristics do you look at when identifying a certain
persons voice? Obviously people with a deeper voice will produce
sounds at a lower frequency than others, but surely you need more than
that.

That process is still under development.

There have been several tacks tried.

One variable is the frequency spectrum of the sound of several spoken
words. This frequency aspect includes the strength of various vowel
sound "formants" which, I believe, are the various harmonics of vowels
that a specific talker's voice (head cavities) favor.

Another variable is the pace of speaking syllables, pauses, etc. So
both frequency and time dimensions are required.

Angelo Campanella

[Adam Chapman]

Post by Adam Chapman
What kind of characteristics do you look at when identifying a certain
persons voice? Obviously people with a deeper voice will produce
sounds at a lower frequency than others, but surely you need more than
that.

        That process is still under development.
         There have been several tacks tried.
        One variable is the frequency spectrum of the sound of several spoken
words. This frequency aspect includes the strength of various vowel
sound "formants" which, I believe, are the various harmonics of vowels
that a specific talker's voice (head cavities) favor.
        Another variable is the pace of speaking syllables, pauses, etc. So
both frequency and time dimensions are required.
                Angelo Campanella

Thanks, I guessed it was pretty under-developed as i could find hardly
any papers on it. What I did find, however, were papers on musical
instrument recognition. I'm thinking about trying to recognise voices
during spoken vowels, which seem to vary more between people and sound
more like instruments. It would be interesting to run a recognition
algorithm during a movie on my pc, and see if it can distiguish
between actor's voices.

[Jerry Wolf]

On Nov 7, 7:10 pm, Adam Chapman

Post by Adam Chapman
Thanks, I guessed it was pretty under-developed as i could find hardly
any papers on it. What I did find, however, were papers on musical
instrument recognition. I'm thinking about trying to recognise voices
during spoken vowels, which seem to vary more between people and sound
more like instruments. It would be interesting to run a recognition
algorithm during a movie on my pc, and see if it can distiguish
between actor's voices.

Au contraire, speaker recognition has been a subject of active
research for at least 40 years. Do a web search on "speaker
recognition" and you'll find lots. See the IEEE Transactions on
Audio, Speech, and Language Processing and its predecessors, and every
ICASSP conference proceedings since 1976 (IEEE Xplore http://ieeexplore.ieee.org/Xplore/
is a great resource).

cheers,
jerry wolf