Skip to content
ModelListedExternal
Browse files

Miotts 0.6b GGUF

GGUF conversion of Aratako/MioTTS-0.6B + Aratako/MioCodec-25Hz-44.1kHz-v2 for CrispASR.

Sourcehuggingface.co/cstr/miotts-0.6b-GGUF/tree/fab3cdbef802fb5f1047708c27378c6e34ec230f
KindOtherBase modelAratako/MioCodec-25Hz-44.1kHz-v2VersionRevision fab3cdbBgrade

Model source

Kind
Other
File size
768 B–1.31 GB
Base model
Aratako/MioCodec-25Hz-44.1kHz-v2
Version
vRevision fab3cdb
Updated
Oct 19, 2026
Downloads
724
Source
Hugging Face

Source description

MioTTS-0.6B GGUF (v2 codec — 44.1kHz)

GGUF conversion of Aratako/MioTTS-0.6B + Aratako/MioCodec-25Hz-44.1kHz-v2 for CrispASR.

Single GGUF containing the LLM (Qwen3, 28 layers, 1024d) and the MioCodec-v2 waveform decoder with UpSampler. Produces 44.1 kHz mono audio.

Files

FileSizeDescription
miotts-0.6b-f16.gguf1.4 GBFull precision (F16 weights)
miotts-0.6b-q8_0.gguf793 MBLLM Q8_0 + codec F16
miotts-0.6b-q4_k.gguf502 MBRecommended — LLM Q4_K + codec F16
tokenizer.json~14 MBQwen3 BPE tokenizer (place next to GGUF)
en_female.emb.gguf<1 KBEnglish female voice preset
en_male.emb.gguf<1 KBEnglish male voice preset
jp_female.emb.gguf<1 KBJapanese female voice preset
jp_male.emb.gguf<1 KBJapanese male voice preset

Usage

crispasr --backend miotts -m miotts-0.6b-q4_k.gguf \
  --voice en_female.emb.gguf \
  --tts "Hello world, how are you today?" \
  -of output

Important: A voice preset (--voice or -emb) is required for good audio quality. Without it, the codec produces unintelligible audio.

The tokenizer.json must be in the same directory as the GGUF file.

Verified ASR Roundtrip

QuantizationSizeInputASR Output
Q4_K+F16502 MB"The quick brown fox jumps over the lazy dog.""The quick brown fox jumps over the lazy dog." ✅
F322.6 GB"Hello world, how are you today?""Hello world, how are you today? What?" ✅
F16 (uniform)1.4 GB—Codec too noisy (SnakeBeta precision) ❌

The mixed quantization (LLM=Q4_K, codec=F16) is critical — uniform quantization degrades the UpSampler's SnakeBeta activation.

Architecture

  • LLM: Qwen3ForCausalLM (28L, 1024d, GQA 16/8, head_dim=128, vocab=164480)
  • Codec: MioCodec-25Hz-44.1kHz-v2 (FSQ → wave_prenet → conv_upsample → ResNet → AdaLN-Zero decoder → ResNet → UpSampler with SnakeBeta → iSTFT)
  • Output: 44.1 kHz mono PCM

License

Apache 2.0 (MioTTS-0.6B is Qwen3-based). MioCodec weights under MIT.

Credits

Provenance and EU AI Act Art. 53 note

  • Upstream model: Aratako/MioTTS-0.6B — published by Aratako.
  • Upstream licence: apache-2.0. This repository redistributes under the same terms; it grants no rights the upstream licence does not.
  • What was done here: format conversion and/or quantisation only (GGUF). No training, no fine-tuning, no merging, no distillation, no change to architecture, vocabulary or capability. Only the numeric representation of the upstream weights differs.
  • Training data: documented — where it is documented at all — by the upstream provider; see the upstream model card. No training data was used, added or selected by this repository.
  • Provider status: under Regulation (EU) 2024/1689 the upstream authors remain the provider of this model. Converting the serialisation format does not make this repository the provider of a new general-purpose AI model, and no such claim is made. Questions about training content, copyright policy or model capability belong upstream.

Sources

1 source
Hugging FacePrimaryVersion fab3cdbef802fb5f1047708c27378c6e34ec230f
huggingface.co/cstr/miotts-0.6b-GGUF/tree/fab3cdbef802fb5f1047708c27378c6e34ec230f

Verified Oct 19

Download files

Model artifacts

7 artifacts

en_female.emb.gguf

gguf · 768 B · Hugging Face

en_male.emb.gguf

gguf · 768 B · Hugging Face

jp_female.emb.gguf

gguf · 768 B · Hugging Face

jp_male.emb.gguf

gguf · 768 B · Hugging Face

miotts-0.6b-f16.gguf

gguf · 1.31 GB · SHA-256 5f7567079a50…f05c · Hugging Face

Download file

miotts-0.6b-q4_k.gguf

gguf · 502 MB · SHA-256 65e3d4692d37…340d · Hugging Face

Download file

miotts-0.6b-q8_0.gguf

gguf · 792 MB · SHA-256 8e2189decab8…b8cc · Hugging Face

Download file

Source excerpts

3 excerpts
ReadmeAug 3huggingface.co/cstr/miotts-0.6b-GGUF

--- language: - en - ja license: apache-2.0 library_name: crispasr base_model: - Aratako/MioTTS-0.6B - Aratako/MioCodec-25Hz-44.1kHz-v2 pipeline_tag: text-to-speech tags: - speech - tts - gguf - crispasr --- # MioTTS-0.6B GGUF (v2 codec — 44.1kHz) GGUF conversion of [Aratako/MioTTS-0.6B](https://huggingface.co/Aratako/MioTTS-0.6B) + [Aratako/MioCodec-25Hz-44.1kHz-v2](https://huggingface.co/Aratako/MioCodec-25Hz-44.1kHz-v2) for [CrispASR](https://github.com/CrispStrobe/CrispASR). Single GGUF containing the LLM (Qwen3, 28 layers, 1024d) and the MioCodec-v2 waveform decoder with UpSampler. Produces **44.1 kHz mono audio**. ## Files | File | Size | Description | |------|------|-------------| | `miotts-0.6b-f16.gguf` | 1.4 GB | Full precision (F16 weights) | | `miotts-0.6b-q8_0.gguf` | 793 MB | LLM Q8_0 + codec F16 | | `miotts-0.6b-q4_k.gguf` | 502 MB | **Recommended** — LLM Q4_K + codec F16 | | `tokenizer.json` | ~14 MB | Qwen3 BPE tokenizer (place next to GGUF) | | `en_female.emb.gguf` | <1 KB | English female voice preset | | `en_male.emb.gguf` | <1 KB | English male voice preset | | `jp_female.emb.gguf` | <1 KB | Japanese female voice preset | | `jp_male.emb.gguf` | <1 KB | Japanese male voice preset | ## Usage ```bash crispasr --backend miotts -m miotts-0.6b-q4_k.gguf \ --voice en_female.emb.gguf \ --tts "Hello world, how are you today?" \ -of output ``` **Important:** A voice preset (`--voice` or `-emb`) is required for good audio quality. Without it, the codec produces unintelligible audio. The tokenizer.json must be in the same directory as the GGUF file. ## Verified ASR Roundtrip | Quantization | Size | Input | ASR Output | |-------------|------|-------|------------| | Q4_K+F16 | 502 MB | "The quick brown fox jumps over the lazy dog." | "The quick brown fox jumps over the lazy dog." ✅ | | F32 | 2.6 GB | "Hello world, how are you today?" | "Hello world, how are you today? What?" ✅ | | F16 (uniform) | 1.4 GB | — | Codec too noisy (SnakeBeta precision) ❌ | The mixed quantization (LLM=Q4_K, codec=F16) is critical — uniform quantization degrades the UpSampler's SnakeBeta activation. ## Architecture - **LLM**: Qwen3ForCausalLM (28L, 1024d, GQA 16/8, head_dim=128, vocab=164480) - **Codec**: MioCodec-25Hz-44.1kHz-v2 (FSQ → wave_prenet → conv_upsample → ResNet → AdaLN-Zero decoder → ResNet → UpSampler with SnakeBeta → iSTFT) - **Output**: 44.1...

NoteOct 19huggingface.co/cstr/miotts-0.6b-GGUF

Source context: 724 downloads · 0 likes · Pipeline text-to-speech · Library crispasr · Repo cstr/miotts-0.6b-GGUF

NoteAug 3huggingface.co/cstr/miotts-0.6b-GGUF

Source context: 540 downloads · 0 likes · Pipeline text-to-speech · Library crispasr · Repo cstr/miotts-0.6b-GGUF