CoolFace
Modelpublic

aufklarer/Indic-Mio-MLX-fp16

sourceHugging Faceapache-2.0updated 3mo agoView on Hugging Face
1likes70downloads
Model Card

Indic-Mio-MLX-fp16

Apple Silicon fp16 export of SPRINGLab/Indic-Mio, bundled with the MioCodec 25 Hz / 24 kHz companion required to decode Indic-Mio content tokens into waveform audio.

This is an artifact bundle for Swift/MLX runtime work. It preserves upstream key names; runtime code is expected to implement model loading, sampling, MioCodec conditioning, and waveform decoding.

Model

FieldValue
Primary sourceSPRINGLab/Indic-Mio
Primary source revision25feace00ca76c71c40b1e8d921fd3c2943c545e
Bundled codec sourceAratako/MioCodec-25Hz-24kHz
Bundled codec revision3a737f0de2c6324cb2fe40c1fbd1056c7add423d
FormatMLX fp16 safetensors
Primary licenseApache-2.0
Codec licenseMIT
Total bundle size1.5 GB
Sample rate24,000 Hz
Voice conditioningMioCodec global embeddings / zero-shot voice path
Runtime statusArtifact export; Swift/MLX runtime port required

Emotion Control

FieldValue
Marker syntaxSuffix angle tags
Supported markers<happy>, <sad>, <angry>, <disgust>, <fear>, <surprise>

Files

FileSizeDescription
README.md2.3 KBModel card.
added_tokens.json309.6 KBTokenizer asset copied from the upstream Indic-Mio repository.
bundle_config.json697 BSoniqo bundle metadata linking the text model and bundled codec.
chat_template.jinja205 BUpstream generation/template metadata.
config.json1.4 KBUpstream Qwen/Indic-Mio model config at repo root for runtime loading and HF download tracking.
generation_config.json183 BUpstream generation/template metadata.
merges.txt1.7 MBTokenizer asset copied from the upstream Indic-Mio repository.
miocodec/config.json230 BMioCodec runtime/export config.
miocodec/config.yaml2.4 KBMioCodec runtime/export config.
miocodec/model.safetensors261.6 MBBundled MioCodec fp16 weights required for waveform decoding.
miocodec/soniqo_manifest.json1.3 KBExport manifest with source revisions, license posture, and bundle metadata.
model.safetensors1.2 GBIndic-Mio fp16 weights with upstream key names preserved.
soniqo_manifest.json3.1 KBExport manifest with source revisions, license posture, and bundle metadata.
special_tokens_map.json613 BTokenizer asset copied from the upstream Indic-Mio repository.
tokenizer.json13.8 MBTokenizer asset copied from the upstream Indic-Mio repository.
tokenizer_config.json2.3 MBTokenizer asset copied from the upstream Indic-Mio repository.
vocab.json2.8 MBTokenizer asset copied from the upstream Indic-Mio repository.

Validation

TestResult
Indic-Mio + MioCodec local roundtripPassed
Input textनमस्ते, मैं आज बहुत खुश हूँ। <happy>
ASR transcriptनमस्ते, मैं आज बहुत खुश हूँ
Sample rate24000 Hz

The roundtrip test was a local synthesis and transcription sanity check. It verifies that the exported Indic-Mio weights and bundled MioCodec path can produce intelligible Hindi speech for an explicit emotion marker. It is not a full runtime benchmark.

Usage

These artifacts are not a drop-in Transformers checkpoint. They are intended for speech-swift or another MLX runtime that implements the Indic-Mio token generation path and MioCodec decoding path.

bash
git lfs install
git clone https://huggingface.co/aufklarer/Indic-Mio-MLX-fp16

Expected bundle layout:

text
Indic-Mio-MLX-fp16/
  model.safetensors
  config.json
  tokenizer.json
  bundle_config.json
  soniqo_manifest.json
  miocodec/
    model.safetensors
    config.json
    config.yaml
    soniqo_manifest.json

Source

  • —Indic-Mio upstream model: https://huggingface.co/SPRINGLab/Indic-Mio
  • —MioCodec upstream model: https://huggingface.co/Aratako/MioCodec-25Hz-24kHz

Links