EveryVoiceTTS / EveryVoiceTTS/EveryVoice

Remove speaker option

Open
#715 0 comments 0 reactions 0 assignees View on GitHub
enhancement
Dominant language
Python
Stars
45
Forks
4
Avg merge
1d 8h
Merged PRs (30d)
14

Description

### Description & Motivation

in addition to renaming a speaker, we should have the ability to delete a speaker which would:

1. remove a particular speaker key from the speaker2id lookup
2. re-initialize weights in the speaker embedding for that speaker id
3. return and save the resulting model to disk

This would allow speaker information to be hidden at the speaker's request.

### Pitch

_No response_

### Alternatives

_No response_

### Additional context

_No response_

Contributor guide

Open the contributing guide

Research direction

Start by locating the speaker2id lookup, the speaker embedding weights, and the model save or return path. Trace how speaker renaming currently works and determine how deletion should update each of these areas. Done means the selected speaker is removed from the lookup, its embedding weights are re-initialized, and the resulting model is returned and saved to disk.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, pytorch
Domain
machine-learning
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.