Limit processing when using pixels-per-second

[xeruf]

When using pixels-per-second, one cannot add --end - understandable, since it is redundant to --width then. But unfortunately, --end is the only flag that limits the time of audio that is processed, so that even on a short width, long audio files take a long time to process with pixels-per-second. Through a simple division this should be limited internally :)

[sgloe]

I am creating a json data file and only need a specific piece of my whole audio file as waveform. I use the --pixels-per-second 15 to reduce the filesize. For the whole file, this works perfectly. Now I tried to only get the part between 00:10 and 00:40 by using --start 10 --end 40. But this fails due with the message Specify either --end or --pixels-per-second but not both. I do not understand why. Is there a way to achieve this?

[chrisn]

Now I tried to only get the part between 00:10 and 00:40 by using --start 10 --end 40. But this fails due with the message Specify either --end or --pixels-per-second but not both. I do not understand why. Is there a way to achieve this?

I think this should be allowed, i.e., set --start, --end, and --pixels-per-second, and use these to determine the width. I'll work on a fix.

As a workaround, you can set the --width option instead of --pixels-per-second. So in your example, you want a 30 second waveform at 15 pixels per second, so the width is 450 pixels: --start 10 --end 40 --width 450.

[xeruf]

It would also be nice to automatically calculate the end from the width. From the behavior I observed when setting width it still processes the audio file till the end and then cuts it off, incurring significant extra processing.

[sgloe]

Hey Chris, I tried your suggested workaround. It does work, if I output a PNG file. The image looks correct and does fit the audio. But as far as I can see, it does not work, if I output a JSON data file. The JSON data always contains the peaks for the complete length of the original audio file. @xeruf what output format do you use?

[chrisn]

Ah, sorry, I should have checked that would actually work before suggesting it. The options are handled differently for creating images vs creating data files, hence the discrepancy - for historical reasons, as we only needed full duration data files, whereas images could be any period within the audio.

[xeruf]

output as image for display with timg

[sgloe]

Ok, I understand. Unfortunately, I am not a CPP Developer, so I am unable to contribute here. But is there a chance, you and your contributors could implement that feature, that enables me to create Waveform JSON Files, the same way as it is possible for PNG Images respecting --start and --end arguments? I can create a dedicated Issue here, if that helps. Otherwise I will find a way to workaround this limitation by using an already shortened WAV File.

[chrisn]

Well, it's just me in my spare time really - but yes. There's no need to create an issue.

[chrisn]

I'm going to look at this together with https://github.com/bbc/audiowaveform/issues/157, as they're related.