Re: [Audacity-devel] Introducing the Internet Archive and Audacity

[Vaughan J. <va...@au...>]

(fwd to -devel list)

Hank Bromley wrote:
> Hi, Vaughan.  Yes, we've been using the Silence Finder, and it's 
> helpful, but has some trouble with this source material (noisy 
> recordings made from vinyl) - it's been difficult to generate a set of 
> dB & length options that work well across LPs, while keeping to a 
> manageable level of false positives and negatives.
>
> As Dan mentioned, we're wondering if we could improve the performance 
> by giving it some hints:  maybe the total number of breaks we expect 
> to find, and/or rough timing indicators of where we expect the breaks 
> to be.  Or maybe another configurable option for how much 
> above-threshold signal to permit within an otherwise below-threshold 
> region before the region is disqualified, so it would be tolerant of 
> brief pops and clicks.
>
> -- Hank Bromley
>
> On Wed, 9 Dec 2009, Vaughan Johnson wrote:
>
>> Thanks, Dan, and thanks again, Robert. I'm cc-ing this to our 
>> developer list. (Audacity -devel participants, please reply all on 
>> this so the IA folks see it. IA team, you'll need to subscribe to the 
>> list to participate, so probably easier to just reply to me.)
>>
>> Have you been using the Analyze > Silence Finder plugin? Is it 
>> problematic?
>>
>> - Vaughan
>>
>>
>> dan wrote:
>>> Rob, thank you so much for the intro.
>>>
>>> Hi Vaughan,
>>>
>>> I've admired Audacity for a long time, and in fact there's a whole 
>>> crew here
>>> at the archive who are interested in Audacity, sound digitization 
>>> and DSP
>>> in general, metadata, track marking, and how it all interacts.  I've 
>>> copied
>>> them just so there's a general awareness that something involving
>>> Audacity is afoot (and i'm probably missing peeps too, because this 
>>> kind of
>>> technical thing mixed in with code we can share around really hits 
>>> the spot here).
>>>
>>> In very general terms, we have long waveforms (from LPs) that are not
>>> divided into tracks, and we'd like to see what we can do with Audacity
>>> to break them up.  We'd like to give Audacity some hints
>>> about where track breaks might be located in a waveform, so that it
>>> doesn't have to look at every sample in making a decision because we
>>> have some rough idea of where the breaks would be.  We'd like to 
>>> give the
>>> hints through the file system, i think, but possibly also through
>>> human intervention in some cases.
>>>
>>> So we're interested on any suggestions or just insights you might share
>>> with us, ranging from buttons to press in a gui, to scripting, to 
>>> anything else.
>>> And we can get a lot more detailed on this of course.
>>>
>>> No matter what, thanks for the work that you (and Dominic and 
>>> everybody)
>>> has done for Audacity, which we use and enjoy at work and at home.
>>>
>>> dan
>>>
>>> Robert Kaye wrote:
>>>> Hi!
>>>>
>>>> Vaughn, it was really good to see you at SF Music Tech on monday. 
>>>> On Tuesday I spent a bit of time at the Internet Archive and 
>>>> talking to them about their cool new project to preserve a whole 
>>>> pile of LPs. As part of their discussions they mentioned that they 
>>>> were using Audacity in the process and that some parts of the 
>>>> Audacity silence finding feature were giving them some grief. I 
>>>> promised to introduce you to the team so that they could pick your 
>>>> brain to improve their digitization process
>>>>
>>>> I've copied Dan from the Internet Archive on this email -- 
>>>> hopefully you can help them out in this cool project.
>>>>
>>>> Dan,
>>>>
>>>> As promised, here is an introduction to Vaughn, a member of the 
>>>> Audacity team. Could you please direct Vaughn to the appropriate 
>>>> person working on the digitization process at the archive?
>>>>
>>>> Thanks!
>>>>
>>>> -- 
>>>>
>>>> --ruaok        The answer to whether or not something is a good 
>>>> idea should not be taken as an indication of whether I want to do it.
>>>>
>>>> Robert Kaye     --     ro...@eo...     --    
>>>> http://mayhem-chaos.net
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>
>>
>