docs: README — add tts-large-fp8/ section + comparison table

README.md

@@ -61,6 +61,23 @@ This is the premium 7B/9B variant of long-form multi-speaker TTS. Microsoft orig
 - Larger backbone → marginally better prosody and speaker consistency at the cost of ~3× weight size and ~3× VRAM
 - **Shorter** max output: 32K context = ~45 min vs the 1.5B's 64K context = ~90 min. The 1.5B is the better choice for long podcasts; the Large is the better choice for short premium-quality dialog.
 
+### About `tts-large-fp8/` — pre-quantized FP8 build
+
+This is a derivative of `tts-large/` with the LM backbone (Qwen2.5-7B layers — the bulk of the parameter mass at ~14 GB bf16) pre-quantized to **Float8WeightOnly** via `torchao.quantization`. The diffusion head and acoustic/semantic tokenizers (~1.5 GB combined) are **left at bf16** because they're numerically sensitive and the savings on those small modules wouldn't matter anyway.
+
+| | `tts-large/` (bf16) | `tts-large-fp8/` (this) |
+|---|---|---|
+| Disk size | ~17.6 GB | **~9 GB** |
+| Working set on GPU | ~20 GB | **~10 GB** |
+| Min GPU | 12 GB (with NF4 runtime cast) | **8 GB native** |
+| Recommended GPU | 24 GB | **12 GB** |
+| Quality | Native | Near-native (FP8 LM weights round-trip cleanly through bf16 compute) |
+| Load time | Faster (no per-tensor cast) | Slightly slower (saved-config reconstruction) |
+
+**How it was made:** loaded `aoi-ot/VibeVoice-Large` via `from_pretrained` with `TorchAoConfig(quant_type=Float8WeightOnlyConfig(), modules_to_not_convert=["diffusion_head", "acoustic_tokenizer", "semantic_tokenizer", "acoustic_connector", "semantic_connector"])`, then `save_pretrained()`. The saved `config.json` carries the quantization spec, so loaders just call `from_pretrained(...)` and the FP8 layers are reconstructed automatically — no special path required.
+
+**When to pick this over `tts-large/`:** if your GPU has 12-16 GB VRAM and you don't want to rely on the runtime auto-quantization fallback. Same workflow, smaller download, no quality regression on most material.
+
 ### About `realtime-0.5b/voices/*.pt`
 
 These are pickled KV-cache dictionaries (output of running each voice prompt through the full model once, then `torch.save`'d). They are **not flat tensor maps** so they cannot be converted to safetensors — they must be loaded with `torch.load(..., weights_only=False)`. Each is 2–7 MB. Microsoft does not publish the acoustic tokenizer that would let users generate new ones, so this set of 25 is the complete preset library.
@@ -75,6 +92,7 @@ This is the legacy research variant of VibeVoice ASR — the one that runs clean
 |---|---|---|---|---|
 | `tts-1.5b` | Text → speech | EN, ZH (multi-speaker) | ~90 min | Up to 4 speakers via `Speaker N:` script tags + voice cloning from per-speaker reference clips |
 | `tts-large` | Text → speech | EN, ZH (multi-speaker) | ~45 min | Same workflow as 1.5B, premium 7B/9B backbone, higher prosody quality |
+| `tts-large-fp8` | Text → speech | EN, ZH (multi-speaker) | ~45 min | Pre-quantized FP8 build of `tts-large` — same workflow, fits 12 GB cards |
 | `asr-7b` | Speech → text | 50+, code-switching | ~60 min | Diarization, timestamps, hotword support via prompt |
 | `realtime-0.5b` | Streaming text → speech | 11 languages (preset-only) | unbounded | ~300 ms first-chunk latency, single speaker |
 
@@ -105,6 +123,7 @@ These mitigations are baked into the released weights and are preserved in this
 |---|---|---|
 | Model weights — 1.5B / ASR / Realtime | [microsoft/VibeVoice-1.5B](https://huggingface.co/microsoft/VibeVoice-1.5B), [microsoft/VibeVoice-ASR](https://huggingface.co/microsoft/VibeVoice-ASR), [microsoft/VibeVoice-Realtime-0.5B](https://huggingface.co/microsoft/VibeVoice-Realtime-0.5B) | MIT |
 | Model weights — 7B Large | [aoi-ot/VibeVoice-Large](https://huggingface.co/aoi-ot/VibeVoice-Large) — community preserve of the now-removed `microsoft/VibeVoice-Large` (uploaded 2025-09-04, one day before Microsoft's pull) | MIT (preserved) |
+| Model weights — 7B Large FP8 | Derivative of `aoi-ot/VibeVoice-Large` — pre-quantized via `torchao.quantization.Float8WeightOnlyConfig` (LM backbone only) | MIT (derivative) |
 | Voice presets | [microsoft/VibeVoice (GitHub)](https://github.com/microsoft/VibeVoice/tree/main/demo/voices/streaming_model) | MIT |
 | Inference code (TTS variants + ASR + Realtime) | [vibevoice-community/VibeVoice](https://github.com/vibevoice-community/VibeVoice) — Microsoft removed `modeling_vibevoice_inference.py` from the original repo on 2025-09-05 | MIT |