First of all: Thank you for this super-slim, reliable and fast implementation of the echoprint server. It's so easy to install and also - at the moment testing with a database of around 200K songs, very fast.
But I've got one problem: I ingested the echonest free database of fingerprints, using the python script that can be found in one of the closed pull requests.
Those songs are recognized very accurate using snippets of 20s recorded (using pyaudio and alsa) via line in. But I have problems with self ingested songs out of my mp3 collection. Sometimes songs I ingested with a quality of around 320kbit/s
Songs I downloaded from iTunes are somehow better recognized (mainly m4a).
Don't know if this is the handling of the server or if it has something to do with codegen (also recognized that codegen generates different codes depending on the platform it runs on).
Does anyone here experience similar problems? Is there some kind of best practice for ingesting songs? Which filetypes etc.?
First of all: Thank you for this super-slim, reliable and fast implementation of the echoprint server. It's so easy to install and also - at the moment testing with a database of around 200K songs, very fast. But I've got one problem: I ingested the echonest free database of fingerprints, using the python script that can be found in one of the closed pull requests. Those songs are recognized very accurate using snippets of 20s recorded (using pyaudio and alsa) via line in. But I have problems with self ingested songs out of my mp3 collection. Sometimes songs I ingested with a quality of around 320kbit/s Songs I downloaded from iTunes are somehow better recognized (mainly m4a). Don't know if this is the handling of the server or if it has something to do with codegen (also recognized that codegen generates different codes depending on the platform it runs on). Does anyone here experience similar problems? Is there some kind of best practice for ingesting songs? Which filetypes etc.?