Thread: [Audacity-devel] Re: [Audacity-help] Audacity for general time series processing?
A free multi-track audio editor and recorder
Brought to you by:
aosiniao
From: Vaughan J. <va...@au...> - 2005-07-17 00:56:51
|
Volker, I haven't seen a response to your email. Sorry for the delay, but I think it's more appropriate for the developer list, so I'm forwarding it there. You may want to join the list. Your idea sounds interesting & we're always happy to have more help. -Vaughan DEWI, V. Riedel wrote: >Hi, > >first I would like to acknowledge this great piece of software which you should be proud >of. >I am, in general more concerned with the analysis of meteorological time series, like >wind speed and direction, which are actually quite similar to sound files. Those data is >sampled mostly every 10 minutes over periods up to about 10 years. Often one has >around 5 to 10 of those time series and wishes to browse them visually. One needs to >cut out error data. Also, one applies filtering like max-min filters or linear functions on >the data or establishes linear relationsships between two channels. This is all quite >similar to digital sound filtering. >As wind energy is currently only a comparably small business, there is no special >software available for this kind of multi-channel time-series analysis of very long files. I >wonder if audiacity could be adapted to such an application. In fact, the only thing really >necessary would be to import a time series file together with the time stamp and display >that correctly in the form YYYY-MM-DD HH:MM:SS in the editor. >Such a file might look like > >2005-01-01 12:00:00 12.3 22.6 >2005-01-01 12:10:00 12.4 24.6 >2005-01-01 12:20:00 11.3 21.6 >2005-01-01 12:30:00 12.3 25.6 >... > >I don't think the actual data import routine would be the problem but it is unclear to me >if the data storage concept of Audiacity would support time stamps. > >Another requirement would be the opportunity to plot several channels together in the >same plot, over each other. And scatter-plots of one channel vs. another would be >required. Do you think this could possibly be included into the Audacity concept? > >I am really interested in joining such a development, as a C++ programmer and time >series analysis expert. What do you think about such a "large time series browser" >audacity version? It could be of interest also to other engineers that must handle many >long time series in a graphical way. One would say there are many such tools like >Matlab, Octave, Scilab, Gnuplot, Origin, Excel,... But non of those has the graphical >capabilities of Audacity and lets you easily scroll and zoom in large time series. >Possibly, one may in some way couple Audacity with Octave or the Gnu Scientific >Library. > >best regards > >Volker > > > |
From: James C. <cr...@in...> - 2005-07-19 20:54:39
|
Volker, I have often thought that Audacity could be adapted into a multi track chart recorder for use in labs. I think a potential big stumbling block to your proposal is if the data items are not evenly spaced in time, or if there are some samples missing. The wave file format assumes uniform sampling intervals, and all samples present. There is a secondary format used in Audacity for the amplification envelopes, which allows sample times to be specified, but it is designed for much lower volumes of data. Have you looked at the code yet, and if so can any pointers we could give you help? For example, producing a wav file from your data would be a first step you could try, along with modifying the status display for date+hh:mm:ss. If it's looking promising at that stage, we can then talk about whether and how we could use some of the other displays you're suggesting in Audio processing. If they are useful for audio, then I might be able to help with some of the new code. Best regards, --James Crook. > DEWI, V. Riedel wrote: > >> Hi, >> >> first I would like to acknowledge this great piece of software which >> you should be proud of. >> I am, in general more concerned with the analysis of meteorological >> time series, like wind speed and direction, which are actually quite >> similar to sound files. Those data is sampled mostly every 10 minutes >> over periods up to about 10 years. Often one has around 5 to 10 of >> those time series and wishes to browse them visually. One needs to cut >> out error data. Also, one applies filtering like max-min filters or >> linear functions on the data or establishes linear relationsships >> between two channels. This is all quite similar to digital sound >> filtering. >> As wind energy is currently only a comparably small business, there is >> no special software available for this kind of multi-channel >> time-series analysis of very long files. I wonder if audiacity could >> be adapted to such an application. In fact, the only thing really >> necessary would be to import a time series file together with the time >> stamp and display that correctly in the form YYYY-MM-DD HH:MM:SS in >> the editor. Such a file might look like >> >> 2005-01-01 12:00:00 12.3 22.6 >> 2005-01-01 12:10:00 12.4 24.6 >> 2005-01-01 12:20:00 11.3 21.6 >> 2005-01-01 12:30:00 12.3 25.6 >> ... >> >> I don't think the actual data import routine would be the problem but >> it is unclear to me if the data storage concept of Audiacity would >> support time stamps. >> >> Another requirement would be the opportunity to plot several channels >> together in the same plot, over each other. And scatter-plots of one >> channel vs. another would be required. Do you think this could >> possibly be included into the Audacity concept? >> >> I am really interested in joining such a development, as a C++ >> programmer and time series analysis expert. What do you think about >> such a "large time series browser" audacity version? It could be of >> interest also to other engineers that must handle many long time >> series in a graphical way. One would say there are many such tools >> like Matlab, Octave, Scilab, Gnuplot, Origin, Excel,... But non of >> those has the graphical capabilities of Audacity and lets you easily >> scroll and zoom in large time series. Possibly, one may in some way >> couple Audacity with Octave or the Gnu Scientific Library. >> >> best regards >> >> Volker |