MOSS-TTSD-v1.0 needs about 15.1 GB of memory at Q8_0 (a 13.0 GB file) with an 8K context window, so it fits on a 16 GB GPU. This repository has 3 quants, from 10.2 GB (Q4_K) to 20.3 GB (BF16).
MOSS‑TTSD‑v1.0: Audio tokenizer and speech‑to‑speech model in GGUF format
MOSS‑TTSD‑v1.0 is an audio‑to‑audio model built on the Cat architecture. It encodes 24 kHz audio into 12.5 Hz tokens and reconstructs speech, music, and sound effects with high fidelity. The GGUF package contains both encoder and decoder for end‑to‑end inference.
Good at
- High‑fidelity audio reconstruction
- Semantic‑rich tokenization for ASR/TTS
- Low‑latency streaming inference
Limits
- No vision support
- Requires recent audio.cpp for bf16 cloning
- Large VRAM needed for full model
Best for
- Audio compression and streaming
- Speech synthesis with speaker tags
- Audio‑based downstream tasks like ASR
Who it’s for. Developers building audio foundation models or applications that need a unified audio tokenizer and speech‑to‑speech pipeline. The model is suitable for research and production use where high‑quality audio encoding and decoding are required.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K | 10.2 GB | 11.9 GB | Too big | Get file → | |
| Q8_0 | 13.0 GB | 15.1 GB | Too big | Get file → | |
| BF16 | 20.3 GB | 23.2 GB | Too big | Get file → |
For 8 GB VRAM, use the Q4_K quantized file (10.2 GB, ~11.9 GB VRAM at 8k context). For 12 GB VRAM, Q4_K fits comfortably. If 16 GB VRAM is available, Q8_0 (13.04 GB, ~15.1 GB VRAM) provides higher precision. BF16 requires 20.28 GB and ~23.2 GB VRAM, suitable only on 24 GB cards.
What GPU do you need for MOSS-TTSD-v1.0?
These figures add the file size, an fp16 KV cache at an 8K context window and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: no quant fits at an 8K context.
- 12 GB VRAM: the best quant that fits is Q4_K (10.2 GB file, about 11.9 GB in use).
- 16 GB VRAM: the best quant that fits is Q8_0 (13.0 GB file, about 15.1 GB in use).
- 24 GB VRAM: the best quant that fits is BF16 (20.3 GB file, about 23.2 GB in use).
Which MOSS-TTSD-v1.0 quant should you choose?
Q8_0 (13.0 GB) is the middle option in this repository. Close to lossless. It costs roughly twice the memory of a 4-bit file, so it suits cases where you have plenty of VRAM.
If memory is tight, the smallest file is Q4_K at 10.2 GB, with the quality trade-offs that come with fewer bits. The usual sweet spot between file size and quality, and the quant most people start with.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about MOSS-TTSD-v1.0
How much VRAM does MOSS-TTSD-v1.0 need?
About 15.1 GB at Q8_0 with an 8K context window. That is the 13.0 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run MOSS-TTSD-v1.0 on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 11.9 GB, so you would need a larger GPU or CPU offloading.
What is the file size of MOSS-TTSD-v1.0 GGUF?
It depends on the quant: Q4_K is 10.2 GB, Q8_0 is 13.0 GB, BF16 is 20.3 GB.
Where can I download MOSS-TTSD-v1.0 GGUF?
The files are on Hugging Face in the christopherthompson81/MOSS-TTSD-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does MOSS-TTSD-v1.0 use?
Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run MOSS‑TTSD on an RTX 3090 with 24 GB VRAM?
Yes. All quantizations fit on 24 GB. Q4_K uses ~12 GB, Q8_0 ~16 GB, and BF16 ~23 GB.
Does the model support multi‑speaker dialogues?
Yes. Speaker tags [S1] and [S2] mark different speakers and can carry reference recordings.
Is the model licensed for commercial use?
It is released under the Apache‑2.0 license, allowing commercial use with attribution.