Update README.md
Browse files
README.md
CHANGED
|
@@ -1,8 +1,57 @@
|
|
| 1 |
---
|
| 2 |
-
license:
|
| 3 |
-
language:
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
-
|
| 8 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language: de
|
| 4 |
+
library_name: transformers
|
| 5 |
+
tags:
|
| 6 |
+
- text-to-speech
|
| 7 |
+
- tts
|
| 8 |
+
- german
|
| 9 |
+
- chatterbox
|
| 10 |
+
- voice-cloning
|
| 11 |
+
- zero-shot
|
| 12 |
+
- merged-model
|
| 13 |
+
---
|
| 14 |
+
|
| 15 |
+
# Kartoffelbox-v0.1_0.65h2: A Merged German Chatterbox-TTS Model
|
| 16 |
+
|
| 17 |
+
## Model Description
|
| 18 |
+
|
| 19 |
+
This repository contains an experimental German Text-to-Speech model created by merging two fine-tuned [Chatterbox-TTS](https://github.com/anotherjesse/Chatterbox-TTS) models.
|
| 20 |
+
|
| 21 |
+
This model is a **hybrid** that combines the characteristics of the well-known German `Kartoffelbox` model with a custom model trained on a specific [male/female] voice. The goal was to fuse the natural pronunciation of "Kartoffelbox" with the unique vocal identity and robustness of an extensively trained custom model.
|
| 22 |
+
|
| 23 |
+
The final model is the result of a weighted-sum merge, using **65% of the weights** from the custom-trained model and 35% from "Kartoffelbox".
|
| 24 |
+
|
| 25 |
+
**Key Features:**
|
| 26 |
+
- **Language:** German
|
| 27 |
+
- **Type:** Hybrid, Merged Model
|
| 28 |
+
- **Capabilities:** High-quality speech synthesis and Zero-Shot Voice Cloning.
|
| 29 |
+
- **Vocal Characteristics:** [Describe what you hear here. E.g., A clear, male voice with a very natural German intonation, sounding less robotic than many standard models.]
|
| 30 |
+
|
| 31 |
+
## Intended Use and Limitations
|
| 32 |
+
|
| 33 |
+
#### Intended Use
|
| 34 |
+
This model is intended for the following use cases:
|
| 35 |
+
- Generating natural-sounding German speech.
|
| 36 |
+
- Experiments in the field of model merging for TTS.
|
| 37 |
+
- As a foundation for further fine-tuning on other German voices.
|
| 38 |
+
|
| 39 |
+
#### Ethical Considerations and Limitations
|
| 40 |
+
(This section can remain as is)
|
| 41 |
+
Speech synthesis and voice cloning technologies carry risks of misuse. This model should **not** be used to:
|
| 42 |
+
- Clone the voices of individuals without their explicit consent.
|
| 43 |
+
- Create misinformation or "deepfakes".
|
| 44 |
+
- Impersonate, deceive, or harass individuals.
|
| 45 |
+
The user of this model is responsible for its ethical use and for complying with all applicable laws.
|
| 46 |
+
|
| 47 |
+
## How to Use the Model
|
| 48 |
+
|
| 49 |
+
Usage is identical to other Chatterbox models. You will need the `chatterbox-tts` library.
|
| 50 |
+
|
| 51 |
+
**1. Installation**
|
| 52 |
+
|
| 53 |
+
```bash
|
| 54 |
+
# Clone the Chatterbox repository and install its dependencies
|
| 55 |
+
git clone https://github.com/anotherjesse/Chatterbox-TTS.git
|
| 56 |
+
cd Chatterbox-TTS
|
| 57 |
+
pip install -e .
|