SatiricalGuru/beatrice-voice-models

🌿 Project Beatrice V2 — Trump Voice Model

Pre-trained Trump Voice Model weights for Project Beatrice V2 — the open-source real-time neural voice conversion engine (<50ms latency).

---

🎙️ Model Files & Downloads (trump/)

File Name	Description	Size	Download Link
beatrice_paraphernalia_data_00005000.toml	Model Configuration & Checkpoint Manifest (Step 5000)	568 B	📥 Download Config
waveform_generator.bin	Neural Vocoder Waveform Generator Weights	7.9 MB	📥 Download Vocoder
phone_extractor.bin	Phoneme Feature Extractor Weights	7.2 MB	📥 Download Extractor
pitch_estimator.bin	Pitch & Frequency Estimator Weights	3.5 MB	📥 Download Pitch Model
embedding_setter.bin	Speaker Embedding Setter Weights	264 KB	📥 Download Embedding Setter
speaker_embeddings.bin	Target Speaker Vector Embeddings	234 KB	📥 Download Speaker Vector

---

⚡ How to Load this Model in Beatrice Voice Changer

1. Download Model Folder

Clone or download the trump/ folder directly:

git lfs install
git clone https://huggingface.co/SatiricalGuru/beatrice-voice-models

2. Load into Beatrice Voice Changer Client

1. Open Beatrice Voice Changer (macOS Metal/MPS or Windows CUDA).
2. Under Model / Target Voice, select Load Directory / TOML and point to the trump/ directory (or select beatrice_paraphernalia_data_00005000.toml).
3. Map your Microphone Input and Virtual Audio Output (BlackHole on macOS, VB-Cable on Windows).
4. Start live real-time voice conversion!

---

🛠️ Performance & Requirements

• Latency: Sub-50ms glass-to-glass real-time inference.
• Hardware Acceleration: Apple Silicon MPS (macOS) / NVIDIA CUDA & DirectML (Windows).
• Total Model Package: ~19 MB complete model footprint.

🌐 Visit the Official Project Beatrice Website →