# User:Nikki/Recording lengths

Since User_talk:Reosarevok/Recording_Issues is really not helping us decide what to do about MBS-2021...

**Note:** This page doesn't refer to standalone recordings. They don't have tracks and therefore don't have track length which means the lengths can only be entered manually. If you want to get rid of standalone recordings entirely, that's something separate from what we're trying to solve here and is a whole new can of worms. Make your own proposal for that if that's what you want.

## Should recording lengths ever be calculated automatically from track lengths?

### Yes - some or all of the time

- Nikki (talk) 10:17, 7 January 2013 (UTC)
- Ianmcorvidae (talk) 10:21, 7 January 2013 (UTC)
- OliverCharles (talk) 10:33, 7 January 2013 (UTC)
- warp
- jacobbrett (talk) 10:46, 7 January 2013 (UTC)
- Murdos (talk)
- Hawke (talk)
- LordSputnik
- monxton
- Hrglgrmpf (talk) 00:40, 13 January 2013 (UTC)
- Freso (talk) 17:59, 13 January 2013 (UTC) (but really earlier than this too...)
- Lukáš Lalinský (talk) 18:02, 13 January 2013 (UTC)
- Mineo (talk) 18:04, 13 January 2013 (UTC)
- reosarevok --Reosarevok (talk) 09:11, 15 January 2013 (UTC)
- Jesus2099 (talk) 09:15, 15 January 2013 (UTC)

### No - editors should always enter them by hand

## Should recording lengths be editable?

### Yes, always

### Yes, but only when there are multiple tracks

### No, always calculate them automatically when there are tracks

- Nikki (talk) 10:17, 7 January 2013 (UTC)
- OliverCharles (talk) 10:33, 7 January 2013 (UTC)
- warp
- Ianmcorvidae (talk) 10:38, 7 January 2013 (UTC)
- jacobbrett (talk) 10:46, 7 January 2013 (UTC)
- Murdos (talk)
- Hawke (talk)
- LordSputnik
- Hrglgrmpf (talk) 00:40, 13 January 2013 (UTC)
- Freso (talk) 17:59, 13 January 2013 (UTC)
- Lukáš Lalinský (talk) 18:02, 13 January 2013 (UTC)
- Mineo (talk) 18:04, 13 January 2013 (UTC)
- Jesus2099 (talk) 09:19, 15 January 2013 (UTC)

## If recording lengths should only be calculated automatically some of the time, when should it happen?

Vote for both if you like, they're not mutually exclusive.

### When there isn't an existing recording length

- monxton
- Freso (talk) 17:59, 13 January 2013 (UTC)
- reosarevok --Reosarevok (talk) 09:11, 15 January 2013 (UTC)

### When the recording only has one track

- monxton
- Freso (talk) 17:59, 13 January 2013 (UTC)
- reosarevok --Reosarevok (talk) 09:11, 15 January 2013 (UTC)

### ...

## If recording lengths should always be calculated automatically from track lengths, how should the length be determined?

### I don't care, just calculate them from track lengths somehow

- Nikki (talk) 10:17, 7 January 2013 (UTC)
- Mineo (talk) 18:04, 13 January 2013 (UTC)
- reosarevok --Reosarevok (talk) 09:11, 15 January 2013 (UTC)

### Sort the releases (e.g. by status then by date) and use the first available length

### Take the mean average of the lengths

### Take the (rounded to avoid decimal durations) median length

- Hawke (talk) — Note that the modal track duration will not work if no track duration is more common than the others, e.g. (4:03, 4:05, 4:06, 4:07)
- Lukáš Lalinský (talk) 18:06, 13 January 2013 (UTC)
- Murdos (talk)

### Take the modal track time

- OliverCharles (talk) 10:33, 7 January 2013 (UTC)
- warp
- I’m slightly partial to my more complex solution jacobbrett (talk) 10:46, 7 January 2013 (UTC)
- LordSputnik (I also think we should be able to select which track time to use, as jacobbrett says. Perhaps use the modal time by default, but allow edits to select the track time to use.)

### Take the shortest length

- Hrglgrmpf (talk) 01:05, 13 January 2013 (UTC)

If we assume the longer lengths are often due to appended/prepended silence, this makes a lot of sense. E.g. if I have a recording with one associated track of 3 min, and two tracks with 3 min + 5 min silence, I want the recording length to be 3 min.

But "modal" is also ok since it is better than the current situation. - Freso (talk) 17:59, 13 January 2013 (UTC) - what hgrl... said.

### ...

## Other views

### Favour track lengths from tracklists which are supported by a DiscID

- Take the mean of track lengths which have a supporting disc ID. If none, do whatever. monxton

### There is no spoon^wrecording length

ocharles thinks that abolishing the concept of recording lengths entirely is the way to go.

- I'm not actually
*quite*sure what you mean by this -- I could see you meaning any combination of: the implementation-concern "don't store this in a database column", don't return it in the webservice, don't display a unified time for recordings, actively discourage people who assert that recordings have an associated time. #4 seems pretty drastic; 2 and 3 are essentially the same thing but for the WS versus the site; and #1 is of course an implementation concern. So, I suppose you mean the combination of #2/#3, which also seems sort of implementation-y -- so, could you clarify exactly what you mean here? It seems like I'm missing the relevance at this stage of the discussion :) Ianmcorvidae (talk) 10:51, 7 January 2013 (UTC) - I agree for recordings on releases, but before we can do this, we need to improve the standalone recording situation. Kepstin (talk) 20:41, 7 January 2013 (UTC)
- I agree entirely. Recordings are probably going to be mixes. We can only really make an educated guess at the mix length, and there's no real use for Mix length when we have track lengths. LordSputnik

### Pull recording durations from acoustIDs, if possible

It’d be nice to have acoustID (or something) as a secondary support for recording durations, since not all releases have a discID (and this will only be more true in the future) —Hawke (talk)

### Durations affected by mix?

Durations are likely to vary more if we’re adjusting to “recording=mix” instead of “recording is somewhere between mix and master”. Will this change anything on how we calculate the duration? There’s also LordSputnik’s WavePlot which apparently has a trimmed length that might be viable. —Hawke (talk)

### Warn for possible incorrect durations

Highlight (red?) durations of tracks where the duration is too far from the auto-detected one (5 seconds? 10 seconds? Something based on track lengths, e.g. *n* standard deviations? —Hawke (talk)