Every week, Cron starts Streamripper to record a programme streamed by my local community radio station. The stream is a continuous mp3.

The station has a fairly liberal idea of when a show begins and ends, so Streamripper starts a couple of minutes before the scheduled start time, and ends a couple of minutes after the end time. I can then use mp3splt to manually remove the excess, so that the resulting file starts and ends exactly at the start and end of the programme. But I have to find these start and end times by playing the file in a media player (or, if I can faff about with it, by using mp3splt-gtk, which is more accurate, but I find fairly clumsy to use).

I was wondering if anyone could think of a way to automate this. I have tried using mp3splt to detect any silence before and after the programme, but it is unreliable, as there isn't always a silence. The programme starts and ends with a theme tune, so maybe something could somehow 'recognise' this specific part of the file, and return the correct time values that mp3splt needs? Or could an inaudible flag be embedded in the theme music that something could detect (I could probably get the programme producer to co-operate with that)?

Is it even possible?

Anything is possible, we humans are the living proof of that...for better...or worse...

Automating an edit? Not likely...yet, so, manually then...

Audacity, Sweep, MHwaveedit to name a few...though there may well be better stuff around...
I personally did'nt use Sweep yet, but my experiences with Audacity and MHwaveedit are satisfactory...

Hope it helps

Thor

Thanks, Thor. I hadn't come across Sweep and MHwaveedit before - I shall enjoy having a play with them.

mp3split does the job I need extremely well; it's not really a question of how to edit an audio file, rather than how to find the times of specific sections in the audio for the edit points.

Audacity is an excellent tool, but not really suited to this job - it imports mp3s, de-coding them into its native format, before it can deal with them, and then has to re-code them back to mp3 when it exports them, which means a loss of quality at each re-code (I would guess the same may be true of Sweep and MHwaveedit). It's probably not too significant in this case, but the real problem is that to load a big mp3 into Audacity can take a long time, and exporting it out again even longer. mp3splt doesn't need to decode, and so it's much faster, and doesn't affect quality. The command in this case is very simple - it's automating the finding of the start and end times that I was wondering about.

I suspect you may be right about it, but some clever person may have figured out a way.