Just a pondered thought...

While doing some recording for a vocalist, I thought of a feature that
could potentially be developed for DAW's that currently doesn't exist.
All DAW's are somewhat 'stupid' when it comes to playback. Wouldn't it
be great if they could recognize the presence of words and identify what
the words were? It must be possible. After all melodyne can accurately
track pitch and rhythm and show the corresponding musical notation. And
God knows that there is a lot dictation software available. Why can't
the two be amalgamated?

In practice this is how it would work:
1. The singer would sing in their scratch from beginning to end (which
usually happens anyway when adjusting pre's, eq's, compressors, etc).
2. The software/plug-in would analyze the passage.
3. The engineer would quickly go over it to check for any obvious errors
in word/pitch detection and correct it.
4. Now when performing the good vocal, an artist can say go back to
where i say, "lucy in the sky..." and the engineer could quickly locate
it either manually or, if it's a lyric heavy tune, do an Apple/Ctrl+F
for the word.

Alternately, if detection of words are not possible due to the
exagerated rhythms found in music and not spoken word, you should still
have the option of manually entering the words after the pitch/rhythm
have been detected, sort of like how you would enter words when scoring
in sibelius.

It may be a little tech-head of me to request such a feature, but i've
found myself in multiple situations where there have been 4 different
singers in a pop group all singing different parts at different times.
Labelling the events just doesn't cut it.

*I would love to have lanes/tracks available where i could see music
notation with the corresponding lyric.*

This feature would also help a great deal when i continue a project that
i didn't have the benefit of observing from the beginning (we've all had
the artist come in with their tracks that they recorded at home who
pretty much expect you to already be familiar with the tune since
they've already worked on it for weeks).

my two cents,

Roach

Mike,
This isn't exactly what you were looking for, but's it's kind of close & pretty amazing.

http://www.soundsonline.com/sophtml/...WORDS=Vocaloid

As a person who can't carry a tune in a bucket, it's a pretty attractive program.
Stu Nevitt


"Mike Rocha" wrote in message ...
Just a pondered thought...

While doing some recording for a vocalist, I thought of a feature that
could potentially be developed for DAW's that currently doesn't exist.
All DAW's are somewhat 'stupid' when it comes to playback. Wouldn't it
be great if they could recognize the presence of words and identify what
the words were? It must be possible. After all melodyne can accurately
track pitch and rhythm and show the corresponding musical notation. And
God knows that there is a lot dictation software available. Why can't
the two be amalgamated?

In practice this is how it would work:
1. The singer would sing in their scratch from beginning to end (which
usually happens anyway when adjusting pre's, eq's, compressors, etc).
2. The software/plug-in would analyze the passage.
3. The engineer would quickly go over it to check for any obvious errors
in word/pitch detection and correct it.
4. Now when performing the good vocal, an artist can say go back to
where i say, "lucy in the sky..." and the engineer could quickly locate
it either manually or, if it's a lyric heavy tune, do an Apple/Ctrl+F
for the word.

Alternately, if detection of words are not possible due to the
exagerated rhythms found in music and not spoken word, you should still
have the option of manually entering the words after the pitch/rhythm
have been detected, sort of like how you would enter words when scoring
in sibelius.

It may be a little tech-head of me to request such a feature, but i've
found myself in multiple situations where there have been 4 different
singers in a pop group all singing different parts at different times.
Labelling the events just doesn't cut it.

*I would love to have lanes/tracks available where i could see music
notation with the corresponding lyric.*

This feature would also help a great deal when i continue a project that
i didn't have the benefit of observing from the beginning (we've all had
the artist come in with their tracks that they recorded at home who
pretty much expect you to already be familiar with the tune since
they've already worked on it for weeks).

my two cents,

Roach

Mike Rocha wrote:
Just a pondered thought...

Wouldn't it
be great if they could recognize the presence of words and identify what
the words were?

If you can find anything, human, computer or otherwise that can
understand what the hell Eddie Vedder is saying I'd love to see it =)

Jonny Durango

That WOULD be cool.... Having tried to use voice dictation
professionally (admittedly last attempt a couple of years ago) my
guess is that it would encumber rather than assist the engineer....

In a quiet room, with a specific mic and geometry, and hours of
training voice recognition software to a specific specialized
vocabulary, I got up to a pretty good, like 90% or so accuracy rate,
by..speaking..slightly.. slower..than..normal. It was still faster
than typing, but not when you account for catchihng the mistakes and
chuckling over the mistakes it makes.... Now imagine a random
speaker, not particularly attempting to be clear for voice
recognition, let alone singing!!!

peter

On Mon, 08 Aug 2005 17:07:50 -0700, Jonny Durango
wrote:

Mike Rocha wrote:
Just a pondered thought...

Wouldn't it
be great if they could recognize the presence of words and identify what
the words were?

If you can find anything, human, computer or otherwise that can
understand what the hell Eddie Vedder is saying I'd love to see it =)

Jonny Durango