Skip to content

Word-by-word timing would be perfect for karaoke #362

@salfter

Description

@salfter

Got this up and running on my home server after reading about it at Noted...even dug up a GeForce GTX 1070 out of storage that I'd previously used for mining to speed it up. I originally figured I might use it to fix subtitles on some of the TV shows in my collection, but then I got the idea to use it for generating karaoke song files. It does a good job separating vocals from lyrics (and both files can be downloaded), and an .srt file is easily massaged into an .lrc file that can then be used as-is or converted to .cdg.

What I'm doing now, though, is going through a song in a karaoke editor to add timing to each word (as opposed to each line). If this program could be set up to kick out timing information for each word, that'd be perfect for this purpose.

Here's a snippet of an .lrc file that originated as an .srt from Whisper-WebUI:

[01:56.48]Never gonna give you up[01:58.40]
[01:58.40]Never gonna let you down[02:00.65]

[02:00.65]Never gonna run around and desert you[02:04.56]
[02:04.56]Never gonna make you cry[02:06.61]

[02:06.61]Never gonna say goodbye[02:08.64]
[02:08.64]Never gonna tell a lie[02:11.10]

[02:11.10]And hurt you[02:12.86]

With some words taking longer than others, per-word timing would be useful. It ends up making something like this (from earlier in the same song):

[00:27.20]We're [00:27.89]no [00:28.21]strangers [00:28.68]to [00:29.17]love[00:31.05]
[00:31.05]You [00:31.79]know [00:32.00]the [00:32.26]rules [00:33.01]and [00:33.28]so [00:33.80]do [00:34.08]I[00:35.12]

[00:35.12]Our [00:35.81]full [00:36.02]commitment's [00:36.83]what [00:37.06]I'm [00:37.64]thinking [00:38.11]of[00:39.28]
[00:39.28]You [00:39.96]wouldn't [00:40.43]get [00:40.67]this [00:40.96]from [00:41.48]any [00:41.95]other [00:42.50]guy[00:43.12]

[00:43.12]I [00:44.77]just [00:45.03]wanna [00:45.53]tell [00:45.84]you [00:46.08]how [00:46.34]I'm [00:46.60]feeling[00:48.23]
[00:48.23]Gotta [00:49.16]make [00:49.39]you [00:49.86]understand[00:50.83]

I produced the timing manually in this block (other than at the start and end of each line). Being able to get this timing automatically would be cool.

(Before you say that the song I'm looking to convert is widely available from the karaoke sites, this is the source I'm using. :) )

Metadata

Metadata

Assignees

Labels

enhancementNew feature or request

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions