Ben-
> I was wondering if anyone had done any direct
comparisons between commercially available products such as Melodyne and
> Hit n Mix which on paper seem to have cracked pitch extraction and polyphonic
recognition and the algorithms presented
> in research journals/conferences etc.
>
> Has anyone run any testing similar to that used at ISMIR? or produced
F-measure scores etc?
>
> I find it interesting that MIR is such an active research area yet commercial
products are coming in to the market
> achieving perceived results which the research community isn't managing
as yet.
That is not surprising. If for-profit companies didn't do substantially
better than academic research groups, then we
wouldn't have an economy, there would be no patent system,etc.
Yes they are better -- that's their job. People invest a lot of money in
those companies, and the engineers work 80
hrs a week (or more) on it for years. Eventually what they're doing may
emerge in the form of published research
papers, company white papers, or other documentation.
-Jeff
> Like i say, this is only perceived as
> I haven't found any direct comparisons as yet.
>
> Has anyone done any testing or comparisons with these products? Or simply have
any comments on the subject of
> commercially products Vs research algorithms?
>
> I'm interested in your thoughts!
> Thanks
>
> Ben.