PennSound Manifesto

1. It must be free and downloadable.

Ideally, all the sound material we put on the web should be cleared for copyright to be distributed free for noncommercial and educational purposes. Users of the site will be able to download the MP3s to their own computers or players or play them in a streaming fashion. Teachers could make course CDs or add the MP3s to their on-line syllabi. Other web sites and libraries could recollect the material. Credits for digitalization and copyright release would also be embedded into each file. One of the advantages of working with poetry sound files is that we don't anticipate a problem with rights. At present and in the conceivable future, there is no profit to be gained by the sale of recorded poetry. There is, however, considerable expense involved in preserving, cataloging, and distributing such material.

2. It must be MP3 or better.

RealAudio is a proprietary format with sound quality that will not stand the test of time. We need to use open formats that reproduce reasonably high quality sound, for a listenership that is used to astoundingly good sound quality from commercial sources.

3. It must be singles.

At present, the vast majority of poetry recordings are for entire readings, typically thirty or more minutes, with no tracking of individual cuts or poems. While these full readings have great literary and archival value taken as a whole, few but the most devoted listen to full recordings of readings or, if they do, fewer still listen more than once. The more useful format is to break readings up into individual poems and to make MP3s of each poem available. MP3s of song-length poems could become a very appealing format for poetry. The implications for audience, listenership, critical thinking, poetics, and poetic production are great.

4. It must be named.

Presently, downloaded poetry sound files tend not to have informative names. Looking at a directory of such files, it is impossible to determine what the file contains from the visual information available. File sharing for music employs a simple system of the name of the singer and the song, but the p2p system is not compatible with FTP, especially in terms of blank spaces between words, which need to have dashes or underlines. In addition, song MP3 file names start with the first name of the singer, but for the poetry files I think we need to move to the more conventional last name first and give the date and place of the performance as well. For poetry file names I propose: lastname-firstname_cut-number_title_place/series_date.mp3.

5. It must embed bibliographic information in the file.

It's important that basic bibliographic information be embedded in the MP3 sound file itself, so that when someone downloads the file they get the right file name and also they get a full range of 'ID3' type information -- all in the same file. This is basically a consumer-oriented MP3 file exchange approach. The goal is to make these sound files available for users to have readily available for play and replay. Since downloaded files will be separated from their home library web site or catalog, information on that web site or in a catalog will not necessarily be retrievable (although the URL for the catalog can also be embedded).

6. It must be indexed.

It must be retrievable both from a library catalog under the author's name and via web search engines.

--Charles Bernstein, 2003