I am trying to automate the maintenance of a directory of Internet Radio streams. I want to get stream bitrate and codec type automatically. Most stations don't put this info in the HTTP header.
I can guess bitrate by streaming a big chunk of data (a meg or more?), timing it, and dividing bits by seconds. If the stream is a true radio station and not a file (aka podcast), this guess should be close to the actual bitrate, modulo some startup burstiness. Unfortunately I'm not yet sure how to tell a true radio station apart from a podcast. The thousand or so stations in my directory all return Content-Type fields that are in this list: text/plain audio/x-mpegurl text/html; charset=ISO-8859-1 audio/x-ms-wax text/html audio/x-scpls text/html; charset=utf-8 audio/x-scpls pls text/plain; charset=iso-8859-1 video/x-ms-asf text/plain; charset=ISO-8859-1 video/x-ms-asx text/plain; charset=UTF-8 video/x-ms-wmx application/octet-stream video/x-ms-wvx application/x-mplayer2 This doesn't give me the details about the codec that I want (is it mp3? flac? wma? encrypted?). So I think I need to get a chunk of data and feed it into some sort of decoder to guess the codec. Can anyone suggest such a decoder/guesser that runs (or can be ported to) linux? Thanks, Dave
