[PD] suggestions for spectral "weight" anaylsis

William Brent william.brent at gmail.com
Fri Jan 3 16:53:41 CET 2014


Oh - and if you're just using one feature, you should probably turn off the
"relative ordering" option with this message to [timbreID]

[relative_ordering 0(


On Fri, Jan 3, 2014 at 10:50 AM, William Brent <william.brent at gmail.com>wrote:

> There are separate versions of each analysis object: one for real time,
> and one for NRT reading straight out of tables. You'll see separate help
> files for [barkSpec~] and [barkSpec], for instance. So an [until] loop
> scanning your pre-recorded audio will be the fastest way for you to work on
> this. That's what's used in the 06/order.pd example. Just look in the [pd
> analysis] sub patch and you can change the feature from barkSpec to
> whatever you like (or whatever combination of features, weighted however).
>
> I'd recommend putting your audio into the timbre-space patch and plotting
> by different features there. That way, you can see how the
> vowels/consonants fall on different axes when using certain features.
> That'll give you some intuition on picking the best feature or combo of
> features.
>
> Last - ordering by timbre is always going to be fuzzy unless you can find
> a one-dimensional feature that reflects the timbre aspect you're after.
> Ordering by multi-dimensional features, you might make a big jump along one
> dimension for one step in your ordering, and then a big jump along a
> different dimension for the next step. You never know how much one
> particular feature is contributing the choice of the next step in the
> ordering. In terms of keeping it relatively intuitive to work with, fewer
> dimensions is better. For speech, I'd recommend trying [specBrightness]
> only, with a boundary frequency of about 2.5kHz. That'll separate the
> high-frequency consonants from the more formanty low-mid vowels. You should
> get a decent continuum with just that one feature.
>
>
>
> On Fri, Jan 3, 2014 at 5:30 AM, João Pais <jmmmpais at googlemail.com> wrote:
>
>>  Hi William and all,
>>
>> I thought there would some relevant things in your library. I'll look
>> into your suggestions later.
>> I don't have a patch that other people can look at, but I can try to
>> explain the context a bit better:
>> - I have a sound of ~40s spoken voice. I'm going to split it in fragments
>> (for now 100ms each) and reorder them
>> - one of the possibilities of reordering the fragments would be to have a
>> "continuous" timbre change in the end. E.g. going from noisy consonants to
>> clean vowels
>> - for the analysis, I guess a mixture of pitch and harmonicity (don't
>> know yet in which order it should be done) would be adequate
>>
>> I noticed your objects work in real time. As the analysis is to be done
>> before the performance, I guess I'll either let the sound play throughout
>> to get the analysis data, or then I'll divide the fragments through x
>> analysis patches, to make it run x times faster.
>>
>> In this case it is spoken voice, but I guess it could by anything else.
>>
>> Best,
>>
>> João
>>
>> Hi João,
>>
>> A measure that would give something near 1.0 for white noise and near 0
>> for a sine wave would be "spectral flatness", which is in the timbreID
>> library. But if you're looking to see how well a spectrum's partials line
>> up harmonically, you won't find that in timbreID yet. One quick option
>> would be to use sigmund~ to get the current pitch, then search the spectrum
>> for the amount of energy in bin ranges related to the expected set of
>> harmonics. Compare that with energy in non-harmonic bins. But then, for
>> things like gongs that sound "pitchy" but have inharmonic spectra, that
>> won't be much help. Depends a lot on what you're trying to do.
>>
>> You *might* find specSpread~ useful, which measures how widely or tightly
>> energy is concentrated around the spectrum's center of gravity. It's in
>> units of Hz though.
>>
>>
>>
>> On Thu, Jan 2, 2014 at 12:38 PM, João Pais <jmmmpais at googlemail.com>wrote:
>>
>>> Hello,
>>>
>>> I wanted to ask if there are any suggestions for spectral "weight"
>>> analysis.
>>> With "weight" I mean a factor which would measure the harmonicity of a
>>> sound - e.g. white noise being 1, and a sinus/silence 0. Surely it exists a
>>> propper word for this already, but I don't know one.
>>>
>>> Is there any external or patch around that does something similar?
>>>
>>> Thanks,
>>>
>>> jmmmp
>>>
>>> _______________________________________________
>>> Pd-list at iem.at mailing list
>>> UNSUBSCRIBE and account-management ->
>>> http://lists.puredata.info/listinfo/pd-list
>>>
>>
>>
>>
>> --
>> William Brent
>> www.williambrent.com
>>
>> “Great minds flock together”
>> Conflations: conversational idiom for the 21st century
>>
>> www.conflations.com
>>
>>
>>
>>
>
>
> --
> William Brent
> www.williambrent.com
>
> “Great minds flock together”
> Conflations: conversational idiom for the 21st century
>
> www.conflations.com
>



-- 
William Brent
www.williambrent.com

“Great minds flock together”
Conflations: conversational idiom for the 21st century

www.conflations.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.puredata.info/pipermail/pd-list/attachments/20140103/e467e078/attachment.htm>


More information about the Pd-list mailing list