Reading wave from memory

There is a fork of audioreader that can read a wave file directly from memory like here https://github.com/jksinton/littlesleeper2/blob/master/littlesleeper.py#L453

So like

print "Processing MP3 queue buffer\n"
        # this is essentially a while-true loop
        for mp3data, timestamp in StreamBufferReader(self.queue_read):
            t1 = time.time() # start time used to calculate loop processing time below

            signaldata = None
            timestamp_str = timestamp.strftime("%Y-%m-%d %H:%M:%S.%f")
            try:
                signaldata = audioread.decode(StringIO(mp3data))
            except (audioread.NoBackendError, audioread.ffdec.ReadTimeoutError):
                print timestamp_str,"\tDecode error. Setting default value."

It is it possible to add a new interface to librosa.open to support this streaming mode? See here about this https://github.com/beetbox/audioread/issues/53

[bmcfee]

Thanks for noting this.

Since #845 (and merging #847) , we’re deprioritizing audioread backends (and hope to drop it entirely at some point). Soundfile already supports this kind of streaming directly, and we have documentation illustrating how to do it here, so I don’t see a real need for adding this functionality. Unless I’m misunderstanding your comment?

[loretoparisi]

@bmcfee yes of course due to patents issues it makes sense. Okay thank you I think using soundfile it is enough for this issue, thank you. The mp3 decoding to wav can be done in a different way.