[LAD] [LAU] from 4 to 400 Hz

pshirkey at boosthardware.com pshirkey at boosthardware.com
Tue Jul 19 01:23:15 UTC 2011


> On 07/19/2011 02:45 AM, pshirkey at boosthardware.com wrote:
>>> On 07/18/2011 04:02 AM, pshirkey at boosthardware.com wrote:
>>>>> On 07/17/2011 10:41 AM, Rustom Mody wrote:
>>>>>> On Sun, Jul 17, 2011 at 1:21 PM, Philipp Überbacher
>>>>>> <hollunder at lavabit.com>wrote:
>>>>>>
>>>>>>> Excerpts from Rustom Mody's message of 2011-07-17 05:33:44 +0200:
>>>>>>>> I am preparing to give a talk on the wider ramifications of music.
>>>>>>>> One of the things I wish to demonstrate is that things that look
>>>>>>> different
>>>>>>>> are merely analogs but at different scales.
>>>>>>>>
>>>>>>>> eg if something vibrates at 400Hz we hear a sound of A-flat. If it
>>>>>>>> 'vibrates' at 4 Hz we hear a beat.
>>>>>>>> In the same analogy a 2 vs 3 poly-rhythm (should?) change to a
>>>>>>>> do-so
>>>>>>> chord.
>>>>>>>> And so on.
>>>>>>>
>>>>>>> I suggest you do some experiments before you give a talk. At 4 Hz
>>>>>>> you
>>>>>>> won't be able to hear anything, you won't even be able to reproduce
>>>>>>> a
>>>>>>> 4 Hz sound with common speakers.
>>>>>>>
>>>>>>
>>>>>> You took me quite literally,  [I did put the vibrate into quotes :-)
>>>>>> ]
>>>>>> Let me spell out the experiment in more detail:
>>>>>> Say I have a rhythm in 4/4 time -- 4 even quarter notes, bar
>>>>>> repeating
>>>>>> every
>>>>>> second played by say a click. [What kind of click I am not very
>>>>>> sure;
>>>>>> sharp
>>>>>> with few harmonics would be best I expect]
>>>>>
>>>>> Exactly. Just take a short audio-sample (aka grain) and trigger it
>>>>> repeatedly. Increase the trigger freq. (aka grain-speed) from 4 Hz ->
>>>>> 400Hz.
>>>>>
>>>>> Search the net for granular-synthesis. Your use-case is not the
>>>>> typical
>>>>> grain-synth application, but the principle is the same.
>>>>>
>>>>>> Now if there were some (realtime) way of sliding the tempo from 1
>>>>>> sec
>>>>>> to
>>>>>> millisec I expect the separate clicks would vanish into a hum at
>>>>>> some
>>>>>> stage.
>>>>>>
>>>>>> This (and other such experiments) is what I want to demo.
>>>>>> Ive started looking at chuck.
>>>>>> How does it compare with puredata?
>>>>>>
>>>>>
>>>>> It's a bit of an apples vs oranges question.
>>>>>
>>>>> the main difference: Chuck you program in text, pure-data you
>>>>> graphically connect "objects" (if you know Max/MSP: pure-data is
>>>>> similar).
>>>>>
>>>>> AFAIK, Chuck does not offer GUI elements - you'll need to implement
>>>>> the
>>>>> slider via OSC or use a "text slider".
>>>>>
>>>>>>>> Is there some kind of software where I can make a 4 Hz beat and
>>>>>>>> pull
>>>>>>>> a
>>>>>>>> slider or a freq text box entry until it sound like a A-flat note?
>>>>>>>
>>>>>>> puredata springs to mind, it's easy to use and has everything you
>>>>>>> need.
>>>>>
>>>>> Indeed. Though chuck, supercollider, csound,... could all do the
>>>>> trick.
>>>>>
>>>>> If you know neither of those. Pure-data is probably the easiest to
>>>>> get
>>>>> started with.
>>>>>
>>>>> http://www.timvets.net/video/grains.php will do what you want with
>>>>> Pd.
>>>>>
>>>>
>>>>
>>>> I'm not sure that does what he wants. He asked for a tool that takes
>>>> an
>>>> existing signal/tone and then down tunes it. What you are suggesting
>>>> creates an emulation of that process but generates a completely new
>>>> signal/tone.
>>>>
>>>> It would achieve a similar sound but is functionally quite a different
>>>> process.
>>>
>>> You are right or course. It's not modeling the desired effect
>>> correctly;
>>> Yet it's close enough and much more robust and convicing for a Demo.
>>>
>>> Actually http://www.sonicvisualiser.org/ may be the tool of choice.
>>> Here's a video where it is used to slow down some Bach so that you can
>>> hear the "beating/pulsing" introduce by equal-temperament tuning:
>>> http://www.youtube.com/user/mcldx#p/a/u/0/uOOhvw89jc4
>>>
>>
>>
>> That's an interesting feature of that tool. I was not aware of that
>> functionality.  It appears to work on the time domain not the
>> amplitude/phase of the signal. I assume as it uses a similar code base
>> to
>> librubber band to achieve that functionality.
>>
>> IIRC,  It is still not quite doing what the original request appeared to
>> be asking for.  i.e. down tuning an existing signal. ( He didn't require
>> time adjustment )
>>
>> Would it just be a case for a pitch shifter with a automation fade curve
>> sone in Ardour/qtractor, etc...?
>>
>
> The SuperCollider one-liner that Bernardo posted is actually brilliant.
>


Sure, IIUC it doesn't take an existing waveform and down tune it. But I
don't know enough about SC to know what that code snippet really does. It
looks like it generates a tone then down tunes it.


--
Patrick Shirkey
Boost Hardware Ltd







More information about the Linux-audio-dev mailing list