[L2Ork-dev] Expression of interest in GSoC 2020

[Jonathan Wilkes]

On Wed, Mar 25, 2020 at 8:20 AM Tsz Kiu Pang <osamupang at gmail.com> wrote:
>
> Hi Jonathan,
>
> Thank you very much for your feedback.
>
> On Tue, 24 Mar 2020 at 12:06, Jonathan Wilkes <jon.w.wilkes at gmail.com> wrote:
>>
>> On Mon, Mar 23, 2020 at 7:58 AM Tsz Kiu Pang <osamupang at gmail.com> wrote:

[...]

>>
>> How about an object based on this misconception mentioned from the project idea:
>>
>> "as the frequency in a digital audio signal approaches Nyquist the
>> accuracy of that signal degrades"
>
>
> This sounds interesting. In my mind this could be achieved by having an FFT (or
> something similar) analysing if the high frequency components reaching a certain
> threshold. If it does, then I can add noise to the samples. Please let me know if you
> have something else in mind.
>

It would help to do a search and read some forums and other articles
on that misconception, which is particularly common. Try to find out what people
believe is happening to reduce the quality at those frequencies and
what it would
actually sound like if it were true. I think there's a specific way in
which those
people believe the signal gets distorted, and I'll be curious to know
if you find that,
too.

>>
>> I think having a handful of objects like this would be good idea--
>> perhaps if you can think of one or two more to add to
>> what you currently have. (You might check some digital audio and
>> recording forums to see if you can find a few more
>> misconceptions to build on.)
>
>
> I will try to come up with a couple of more ideas before submitting the proposal
> in the next few days.

Sounds good.

>
>>
>> Also, you should think about how you can take this idea and really
>> refine each object over the course of the
>> summer so that the objects are generally useful. I think your "warm"
>> vinyl idea is a good start-- by the end of the
>> summer it should end up as something more than a rudimentary filter.
>>
>> Does that sound like something which you would be able to successfully complete?
>
>
> Yes it does sound like something that is achievable to me. I spent some time to
> implement the vinyl~ filter. However once I got a hold on how to build the externals,
> I believe it would not be too hard to implements other filters. I guess at this stage, my
> concern would be that it might take me another month from now to get a hold on
> debugging the externals, but even then, it does not sound too hard to me.

That's good to know. I would strongly suggest to spend some time thinking about
what skills you want to acquire over the next month with regard to debugging
external signal classes. What problems do you think you'll run into?
What kind of
tooling would help you be able to debug the code?

If you put those details into your proposal it will help us know how to help you
get up and running as a developer. It can also potentially show us places where
our tooling is currently weak. (For example, I can tell you right now
there is not
a standard regression testing suite for signal classes in Pd atm.)

>
>> >> Then you feed the output of some patch through that filter and let
>> >> someone listen who believes the
>> >> misconception. If the implementation was a success, they should reply
>> >> that the output indeed
>> >> sounds "digital" and "less warm than vinyl." Then you can remove the
>> >> filter to show them that it
>> >> was a trick/fake news.
>> >
>> >
>> > To get feedback from "someone", I guess you are referring to the purr data community?
>>
>> No. Someone who believes the misconception would be a person who
>> believes, say, that
>> analog synths always sound "warmer" and "more present" than digital synths.
>
>
> I am just wondering how would I get those people to listen to my implementation
> of the fake news filters?

I just meant to think about the design of the classes by imagining
such a situation.
But there are many forums online where people discuss DSP, audio filters, and
audio systems. You could potentially demo your implementation for some of those
groups or even get some feedback from them about some of these misconceptions.

>
>>
>> Let me know if you have any more questions. Stay safe and sane during
>> this pandemic!
>
>
> I do not have more questions regarding the fake news library at this stage.
> However I am also interested in the "Data Over Audio Messaging" project
> (sorry if it is a little bit too late to express my interest.)
> Regarding this project, I guess I have two concerns/challenges:

Okay. Just to let you know-- I haven't done research in this area so
the topic is fairly new to me.

> 1.  How should I cancel the feedback such that the sender does not receive its own message?

Good question. Perhaps you could have sender and receiver listening on
different frequencies so that the feedback doesn't matter.

I do know there is some prior art for this. I believe there's even a
Javascript/Web Audio API library that has been implemented to achieve
this.

I also remember seeing some research about sending information through
audio with sounds that audibly pleasing to the listener. If you'd like
I can try to find the link to that.

> 2.  How should I encode the message? Shall I follow some sort of existing protocol (such as TCP or Ethernet Frame or something else) and translate it to an audio equivalent?

TCP seems hard to beat. I'd have a look at as many outstanding
implementations you can find and see how they do it.

>
> Although I have been working on the fake news library project, I believe the skills
> are transferable between these two projects, and that my implementation of the
> vinyl~ filter has shown my basic understanding of the pure data API.

So are you thinking of switching to the "Data Over Audio Messaging" project?

>
> I hope you are all staying safe and healthy.

Same to you! Stay safe.

Best,
Jonathan

>
> Kind regards,
> Tsz Kiu
> _______________________________________________
> L2Ork-dev mailing list
> L2Ork-dev at disis.music.vt.edu
> https://disis.music.vt.edu/listinfo/l2ork-dev