Comments (1)
This is an awesome project, btw. Thanks!
As for being able to limit the amount of writes to disk. Finetuning draws on the wider Python GPT trainer scripts, or to put it another way, scripts I have no direct access to or control of. There's no direct way within Python for me to force those other scripts to behave in a certain way or store their work in memory vs disk.
I can suggest a sort of solution for you though, its a few extra steps mind, but as far as I can think it should keep most (possibly all) activity occurring within RAM. This would be to setup a RAM-Disk and run the training from that. Obviously I've not tried this, so I'm giving you a loose hypothetical on how to do it.
In short a RAM-Disk creates a virtual disk drive, but stored only within your System RAM, meaning if you switch your computer off, that data is gone! Here is some example software (found with a quick Google search) https://www.softperfect.com/products/ramdisk/ I'm sure there are others out there. Microsoft no longer provide a RAM-Disk option as part of the Windows OS (which I believe you use based on your earlier screenshots).
So what you would do is something like this:
- Setup a RAM-Disk (lets say its the K drive).
- Start the Python environment you use.
- Copy the AllTalk folder from its actual C drive location (or wherever) to your K drive RAM-Disk.
- In your command prompt, go to the K:\alltalk_tts folder and run
python finetune.py
- Go through all your finetuning as normal.
- At the end, when it copies the model and sample wav's to K:\alltalk_tts\models\trainedmodel (or whichever one you use) you would need to copy that off the K:\ drive and back to the correct location on your C:\ drive.
- Close down your RAM-Disk.
I am making an assumption here that the majority of disk activity you are seeing is specifically to do with the model, its training and the training data being accessed and written to. I expect there may be an element of the OS still doing something with temporary files, but you cant avoid everything.
So a few extra steps, but should get you what you want.
Hope that gives you solution!
Thanks
from alltalk_tts.
Related Issues (20)
- When installed under webui inside docker, the webserver doesn't start HOT 7
- chinese language is not working HOT 7
- What's going on here? HOT 1
- Move tts_chunk_player.html from /static to a template render HOT 3
- ImportError: cannot import name 'SampleOutput' from 'transformers.generation.utils' HOT 1
- Unable to install, it tries to build something which fails due to missing 'basetsd.h' HOT 7
- 🔄 ImportError: cannot import name 'SampleOutput' from 'transformers.generation.utils'
- start_alltalk.bat window closes instantly on boot HOT 3
- Koboldcpp added XTTS api support HOT 21
- Bat Directly closing HOT 2
- how do i enable sharing? is there a --listen or --api for alltalk tts? HOT 5
- Hi, I use colab for fine tune models but I run AllTalk locally. How can I add new models? I created a folder named ''trainedmodels'' but the UI doesn't recognize it HOT 1
- I fine tuned someone voice. Is there any method/trick to make speech faster, happier, sad, excited, scared ? HOT 1
- Hello, is there any way to do the voices laugh? Is any word or character is equivalent to laught? HOT 1
- Error with Polish characters HOT 3
- [feature request] Narration speed as a setting HOT 3
- Narrator options not showing up in standalone install HOT 2
- Export to Wav in TTS Generator not working HOT 7
- 2 Bugs. Can't set output path and can't change server IP in Sillytavern HOT 3
- ability to use models other than xtts? HOT 4
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from alltalk_tts.