dhruv - we keep the analysis for a very long time. However, the URL that we return in the track/profile call will expire after 10 minutes. Just re-call the track/profile with your track ID to refresh the URL and you should again be able to get the full analysis. -- Paul
↧
How long is analysis stored on echonest servers
↧
How long is analysis stored on echonest servers
Ah sorry for the false alarm. There was a typo in the tracking ID i was using; it wasn't the same as the one EN returned before.
↧
↧
Having trouble Getting Darkside (artist) foreign_ids for musicbrainz and discogs
I working on an app and I am having trouble getting Darkside (artist) foreign_ids for musicbrainz and discogs. I am using the following end point:
http://developer.echonest.com/api/v4/artist/search?api_key=XXXXXXXXXXXX&bucket=id%3Adiscogs&format=json&name=darkside&results=100
The following should be included in the response: http://musicbrainz.org/artist/116b79ab-b049-4299-8fd8-17ac8a18b7f3
http://www.discogs.com/artist/2522005-DARKSIDE-22
Not sure if there is a way I can update this data.
Thanks for your help.
↧
Music identification for PC
yeah ^^
curl -F "api_key=XXX" -F "filetype=mp3" -F "track=@audio.mp3""http://developer.echonest.com/api/v4/track/upload"
is perfect, I tried with a 25 seconds recorded audio and it worked !! thank you
a question : do I have to change api_key? and don't know exactly what it is.
↧
Music identification for PC
don't pay attention to the previous post, I made a mistake to test an audio file with tags info in the propoerties
↧
↧
Static playlist request not returning enough results
Hi --
The following request: http://developer.echonest.com/api/v4/playlist/static?api_key={shhh.....}&format=json&results=100&artist=Jazz+Gillum&bucket=id%3Aspotify-WW&bucket=tracks
consistently returns back 24 results. There are several more available when searching through Spotify directly -- 80 at least. Is that the expected result?
↧
Music identification for PC
Amir,
The track/profile you get back after calling track/upload includes the metadata tags from the audio if no song is identified. If the audio is successfully identified, there will be an Echo Nest SOng ID in the track profile, and the artist and title will be replaced with those of the song.
25 seconds is not quite enough content to get a match; there should be at least 30 seconds. And too much distortion from recording will hamper the identification as well.
David
↧
Cannot get a song to match with identify GET
god to,
Can you send the full URL (including code string) you are sending (but without your API key)?
When you say you "get the full identification response in JSON" without any version parameter, what does that response look like? Does it match a song?
Also please join the Echoprint Google Group for discussion and assistance with Echoprint.
David
↧
getting playlist via years
hello,
newbie here. i have a question or future request. can i create a playlist that contains songs between specific years for example 1925 and 1930. is it possible and how?
thanks
↧
↧
getting playlist via years
Hi Woebegone- it is currently not possible. However, you can restrict a playlist based upon the years active dates for an artist. Here's an example:
http://developer.echonest.com/api/v4/playlist/static?api_key=YOUR_API_KEY&artist_start_year_after=1929&artist_start_year_before=1940&format=json&results=20&type=genre-radio&genre=jazz
Paul
↧
Cannot get a song to match with identify GET
David - thanks for responding. I can no longer reproduce this problem.
The following URL and code parameter data (without a version number) ....
http://developer.echonest.com/api/v4/song/identify?api_key=
... returns the empty songs array: {
"response": {
"status": {
"version": "4.2",
"code": 0,
"message": "Success"
},
"songs": [ ]
}
}
When I add in '&version=4.11' or 4.12 (or 4.
{
"response": {
"status": {
"version": "4.2",
"code": 0,
"message": "Success"
},
"songs": [
{
"score": 31,
"title": "Accidents Will Happen",
"message": "OK (match type 8)",
"artist_id": "AROCZUC1187B9AD05B",
"artist_name": "Elvis Costello & The Attractions",
"id": "SOADJAU12A8C13C656"
}
]
}
}
I am not sure why this behaved as I noted previously, but the problem has gone away.
↧
codegen: Support libav toolset
Hello,
I am running debian so I do not have ffmpeg. Is there any reason why avconv cannot be used in place of ffmpeg? I am more than willing to come up with the ffmpeg -> avconv equivalents for you. What is the `ffmpeg invocation that is used by codegen?
↧
codegen: Support libav toolset
codegen is just an example of using libcodegen. libcodegen's Codegen() entry takes a buffer of PCM float32s. You can generate those samples using anything you want.
We shell out ffmpeg in the example codegen because it's the most popular converter but anyone can use anything they want in their own code.
↧
↧
Loudness - negative decibels
Sorry to reopen this post, but are they real decibels? Like on a logarithmic scale?
I am asking because i want to average them over a year, and in the case of DB I have to transform them first.
↧
Loudness - negative decibels
Hi Dorien - our loudness scale is the subjective judgment of the intensity of a sound. See Tristan's thesis for a detailed description:
http://web.media.mit.edu/~tristan/phd/dissertation/chapter3.html#x1-340003.2
You can treat it as a logarithmic DB scale, but it is not the traditional audio engineers view of DB.
- Paul
↧
Loudness - negative decibels
Thanks, without having read the useful link you sent me. You think it would be very wrong to average it linearly?
↧
Loudness - negative decibels
dorien - it depends on what you want to do with the results, but I think it'd be best to not work in the log domain. -- P
↧
↧
Loudness - negative decibels
I figured so much :/
Oh and then probably, the timbre and other aspects of the PCA are non-linear as well?
Sorry, I am a bit new with the frequency domain.
Thanks for your time btw Paul!
↧
Loudness - negative decibels
Tristan's thesis covers much of the basics. FOr instance, Timbre is covered in section 3.3 -- Paul
↧
Loudness - negative decibels
Hi Paul,
I have a follow-up question. When doing timbre analysis, I get 12 unbounded values (attributes) where the first one represents the average loudness of the segment. (1) Is this first attribute the same as the loudness I get when using track_info['loudness']? (2) If yes, why am I always getting positive values (from 0 up to n)? You mentioned before that loudness values typically range from -60 to 0 (see chart: link text). Could you please clarify this for me?
thanks, John
↧