Use of filename (vs. file-like object) about python-audio-tools HOT 3 OPEN

Here is a very simple implementation I was experimenting with: https://gist.github.com/davidblewett/7340369 . And using it: https://gist.github.com/davidblewett/7340385 . In my case, I will be also writing a web service to extract the necessary metadata to load the file (channels, sample rate, etc). I have to use the pure Python implementation of the FLAC decoder though, since the C one only looks at files on disk. The performance is acceptable, albeit high CPU usage (60-75% on my i7 Macbook Pro; ~30MB memory usage).

What would really make this easier is an abstraction of the object providing the FLAC data, combined with ways to set the necessary metadata attributes without interacting with the actual FLAC data.

from python-audio-tools.

I've just about finished overhauling the FlacDecoder object to take either files or file-like objects (with read() and close() methods) and decode them properly. Only actual on-disk files will support the FlacDecoder.seek() method, but everything else should work okay.

It's essentially changing from FlacDecoder(filename, channel_mask, [bytes_to_skip]) to just FlacDecoder(file) where channel mask will be determined by the decoder and skipped bytes (for ID3v2 tags) will be applied at the Python level.

from python-audio-tools.

I have also been trying to adapt audiotools to work with streaming media. My use case is to decode a podcast for transcription. My fork is at https://sourceforge.net/u/petebleackley/audiotools/ and I would welcome feedback on my efforts so far.

from python-audio-tools.