Hoi, I read this in digest mode so let me answer things together.
The reason why .ogg files are not great is because indeed it is a lossy algorithm. There is some great software to analyse pronunciation files; a program called "praat" is worth mentioning it is even licensed under GPL. There is even functionality in there to do with IPA transcription.
Gregory's proposal to use Ogg/FLAC is not helpfull. This is not the format that is used to analyse pronunciation files. The notion that a specific quality was "the gold standard" at the time is indeed that. It used to be, times have changed.
The Shtooka program that we are talking about CAN create both a WAV and an OGG file. It just needs asking. It would be helpful if we learn sooner rather than later what the outcome is of this request.
Thanks, GerardM
On 2/11/07, Gerard Meijssen gerard.meijssen@gmail.com wrote:
Hoi, I read this in digest mode so let me answer things together.
The reason why .ogg files are not great is because indeed it is a lossy algorithm. There is some great software to analyse pronunciation files; a program called "praat" is worth mentioning it is even licensed under GPL. There is even functionality in there to do with IPA transcription.
Gregory's proposal to use Ogg/FLAC is not helpfull. This is not the format that is used to analyse pronunciation files. The notion that a specific quality was "the gold standard" at the time is indeed that. It used to be, times have changed.
The Shtooka program that we are talking about CAN create both a WAV and an OGG file. It just needs asking. It would be helpful if we learn sooner rather than later what the outcome is of this request.
The Ogg/Flac is lossless, so it removes your concerns about lossyness. It can be uploaded today, so it removes the problems of not being uploadable. It is compressed (losslessly) so it's not quite so bad on our storage and bandwidth. Shtooka already outputs Flac, and could be trivially altered to output ogg/flac, if you'd like I will do this for you. Any number of Ogg/Flac files can be quickly converted to wav with a single command.
I am very hesitant and concerned about the prospects of permitting uncompressed files: I think people will use them where they are completely inappropriate because they are a bit easier to playback. Flac or Ogg/Flac should be substantially smaller than wav and won't drive people to use uncompressed formats for bad reasons.
Hoi, The problem is that existing academic software like "praat" use .wav files. I do sympathise up to a point that storage is used. However, the price of a terabyte of storage is such that this is not that relevant. Both an .ogg and a .wav file would be saved. The first is to enable science to do its thing, the second is for our punters. Thanks, GerardM
http://www.fon.hum.uva.nl/praat/
Gregory Maxwell schreef:
On 2/11/07, Gerard Meijssen gerard.meijssen@gmail.com wrote:
Hoi, I read this in digest mode so let me answer things together.
The reason why .ogg files are not great is because indeed it is a lossy algorithm. There is some great software to analyse pronunciation files; a program called "praat" is worth mentioning it is even licensed under GPL. There is even functionality in there to do with IPA transcription.
Gregory's proposal to use Ogg/FLAC is not helpfull. This is not the format that is used to analyse pronunciation files. The notion that a specific quality was "the gold standard" at the time is indeed that. It used to be, times have changed.
The Shtooka program that we are talking about CAN create both a WAV and an OGG file. It just needs asking. It would be helpful if we learn sooner rather than later what the outcome is of this request.
The Ogg/Flac is lossless, so it removes your concerns about lossyness. It can be uploaded today, so it removes the problems of not being uploadable. It is compressed (losslessly) so it's not quite so bad on our storage and bandwidth. Shtooka already outputs Flac, and could be trivially altered to output ogg/flac, if you'd like I will do this for you. Any number of Ogg/Flac files can be quickly converted to wav with a single command.
I am very hesitant and concerned about the prospects of permitting uncompressed files: I think people will use them where they are completely inappropriate because they are a bit easier to playback. Flac or Ogg/Flac should be substantially smaller than wav and won't drive people to use uncompressed formats for bad reasons.
Hi guys,
I have no knowledge about this software, but Gerard indicated that 'praat' is GPL licensed and its source is available. The code was written in C, I assume, as C99 is required to compile it.
Wouldn't benifits be much greater is support for other file formats in praat was arranged for somehow?
Cheers!
Siebrand
-----Oorspronkelijk bericht----- Van: commons-l-bounces@lists.wikimedia.org [mailto:commons-l-bounces@lists.wikimedia.org] Namens Gerard Meijssen Verzonden: zondag 11 februari 2007 17:55 Aan: Gregory Maxwell CC: Wikimedia Commons Discussion List; The Wiktionary (http://www.wiktionary.org) mailing list Onderwerp: Re: [Commons-l] Sound files
Hoi, The problem is that existing academic software like "praat" use .wav files. I do sympathise up to a point that storage is used. However, the price of a terabyte of storage is such that this is not that relevant. Both an .ogg and a .wav file would be saved. The first is to enable science to do its thing, the second is for our punters. Thanks, GerardM
http://www.fon.hum.uva.nl/praat/
Gregory Maxwell schreef:
On 2/11/07, Gerard Meijssen gerard.meijssen@gmail.com wrote:
Hoi, I read this in digest mode so let me answer things together.
The reason why .ogg files are not great is because indeed it is a lossy algorithm. There is some great software to analyse pronunciation files; a program called "praat" is worth mentioning it is even licensed under GPL. There is even functionality in there to do
with IPA transcription.
Gregory's proposal to use Ogg/FLAC is not helpfull. This is not the format that is used to analyse pronunciation files. The notion that a specific quality was "the gold standard" at the time is indeed that. It used to be, times have changed.
The Shtooka program that we are talking about CAN create both a WAV and an OGG file. It just needs asking. It would be helpful if we learn sooner rather than later what the outcome is of this request.
The Ogg/Flac is lossless, so it removes your concerns about lossyness. It can be uploaded today, so it removes the problems of not being uploadable. It is compressed (losslessly) so it's not quite so bad on our storage and bandwidth. Shtooka already outputs Flac, and could be trivially altered to output ogg/flac, if you'd like I will do this for you. Any number of Ogg/Flac files can be quickly converted to wav with a single command.
I am very hesitant and concerned about the prospects of permitting uncompressed files: I think people will use them where they are completely inappropriate because they are a bit easier to playback. Flac or Ogg/Flac should be substantially smaller than wav and won't drive people to use uncompressed formats for bad reasons.
_______________________________________________ Commons-l mailing list Commons-l@lists.wikimedia.org http://lists.wikimedia.org/mailman/listinfo/commons-l
On 2/11/07, Gerard Meijssen gerard.meijssen@gmail.com wrote:
Hoi, The problem is that existing academic software like "praat" use .wav files. I do sympathise up to a point that storage is used. However, the price of a terabyte of storage is such that this is not that relevant. Both an .ogg and a .wav file would be saved. The first is to enable science to do its thing, the second is for our punters.
I'm quite confidence that anyone using praat can type "flac -d *.ogg" to decode all their files before starting pratt the first time.
In any case, praat should be enhanced to work with Ogg/Flac files. A frequent problem with wavs in this application space is they have no metadata. Ogg/Flac and Flac would be good independent of the advantages of compression. I'm looking at the source right now, and it appears that it would only need a few minor changes to get in basic flac support.