Closed sertansenturk closed 9 years ago
Yes, I guess we can do this. So we need to set the track artist to [dialogue] and then also add the artist back in as a vocal relationship, right?
We need to set the track artist to [dialogue], yes.
However we shouldn't add the artist as a vocal relationship because the tracks are 5-6 second announcement done by somebody else than the artist. Unfortunately I haven't seen any indication of who the presenter is.
Ah, that's fine. We'll just do the [dialogue] then
On 10 March 2015 at 15:59, Sertan Senturk notifications@github.com wrote:
We need to set the track artist to [dialogue], yes.
However we shouldn't add the artist as a vocal relationship because the tracks are 5-6 second announcement done by somebody else than the artist. Unfortunately I haven't seen any indication of who the presenter is.
— Reply to this email directly or view it on GitHub https://github.com/MTG/dunya/issues/195#issuecomment-78071360.
Yes, that's exactly what we need.
Thanks!
The database I have in Dunya doesn't understand track level artist relationships. We'll need to work out where and how to store this. Maybe with a [x] dialogue boolean flag on recordings.
On 10 March 2015 at 16:02, Sertan Senturk notifications@github.com wrote:
Yes, that's exactly what we need.
Thanks!
— Reply to this email directly or view it on GitHub https://github.com/MTG/dunya/issues/195#issuecomment-78072259.
Ok...
Another thing is these recordings are not very interesting since they are not music and hence we don't have to do any analysis on them. Probably such a boolean is also useful so that we don't extract and erroneously show and features (i.e. predominant melody) on them.
Maybe we should make this flag more broad - e.g. analyse/don't analyse. Are there any other kinds of tracks that we don't want to analyse?
On 10 March 2015 at 16:13, Sertan Senturk notifications@github.com wrote:
Ok...
Another thing is these recordings are not very interesting since they are not music and hence we don't have to do any analysis on them. Probably such a boolean is also useful so that we don't extract and erroneously show and features (i.e. predominant melody) on them.
— Reply to this email directly or view it on GitHub https://github.com/MTG/dunya/issues/195#issuecomment-78074615.
Yes, makes sense.
I cannot think of another example right now for makams, but other cultures might probably have cases. Maybe we should open an separate issue for analyse/don't analyse and mention everyone to comment?
After talking about this with andres - The docserver has no knowledge about cultural context. It only knows how to store data. What's more, sometimes we might want to run some analysis tools (image generation), but not others (pitch extraction). Our suggestion is that we process all files, regardless of if they are audio or not. Then we can keep this cultural-specific knowledge in the Makam module, and choose to show the analysis or not.
We will have a [x] don't analyse flag on Makam.recording, and set it on import based on the recording artist.
@sertansenturk, I was working on the script for modifying a recording, when has the name [anon] I have to set the artist to [dialogue]. Do you want to change the name of the recording too or just the artist?
I checked some cover arts. The tracks are typically entered as "Anons." Here's an example. So I think it's better that the recording names stays as is...
Andres has made these edits. They have 2 days to finish. We've also configured the importer to mark these tracks as "don't analyse", so we can choose how to show them in the web frontend. I will leave this ticket open until the edits have passed and we have reimported.
All edits have finished, and this is importing now. It will be finished in 3-4 hours.
We currently don't expose this analyse/don't analyse
flag through the API, Although you could guess this if you look at the recording artist. @sertansenturk if you want this feature, re-open this ticket.
In the corpus we have a lot of tracks mostly from the bootleg which are named as [Anons]. These are very short tracks, which basically have the dialogue announces the next song. Under the MusicBrainz style page, it is explained that the artist should be labeled as dialogue.
I have checked the first page of the recordings with the names Anons. They are all this type and actually the announcer is not the vocalist, it's someone else.
Long story short, can we do automatically convert the artist names in the {Anons} tracks to [dialogue]?