Skip to content
ModelListedExternal
Download

Lyric Align Mms Fa

ONNX export of Meta's MMSFA (Massively Multilingual Speech Forced Alignment) model for CTC forced alignment.

Sourcehuggingface.co/xycld/lyric-align-mms-fa/tree/06e6cbfa5c257d23981c922f6406f5def958eb95
KindOtherVersionRevision 06e6cbf

Model source

Kind
Other
File size
3.10–302 MB
Version
vRevision 06e6cbf
Updated
Sep 13, 2026
Downloads
0
Source
Hugging Face

Source description

MMS Forced Alignment — ONNX

ONNX export of Meta's MMS_FA (Massively Multilingual Speech Forced Alignment) model for CTC forced alignment.

Files

FileSizeDescription
mms_fa.onnx3.2 MBONNX model graph
mms_fa.onnx.data1.2 GBExternal weight data (FP32)
tokenizer.json311 BToken-to-ID mapping (29 tokens)

Usage

from huggingface_hub import hf_hub_download

model_path = hf_hub_download("xycld/lyric-align-mms-fa", "mms_fa.onnx")
data_path = hf_hub_download("xycld/lyric-align-mms-fa", "mms_fa.onnx.data")
tokenizer_path = hf_hub_download("xycld/lyric-align-mms-fa", "tokenizer.json")

Note: mms_fa.onnx.data must be in the same directory as mms_fa.onnx for ONNX Runtime to load correctly. hf_hub_download handles this automatically via its cache.

Model Details

  • Original model: MMS_FA by Meta Research
  • Architecture: wav2vec2-based CTC forced aligner (315M parameters)
  • Precision: FP32
  • Input: mono 16kHz waveform
  • Output: log-probability emission matrix [num_frames, 29 labels] at 50fps (20ms/frame)
  • ONNX opset: 18

Quantized Variants

VariantSizeCompressionLoad TimeInferenceMAEAcc @50msAcc @100msAcc @200msStatus
FP32 (original)1,207 MB1.0x989ms424ms/line34.7ms86.2%97.5%99.4%Available
FP16605 MB2.0x2,335ms576ms/line34.7ms86.2%97.5%99.4%Not recommended
UINT8303 MB4.0x412ms262ms/line34.9ms86.2%97.5%99.4%Recommended

Benchmark: Chinese song "错位时空" (362 characters, 53 lines) on CPU.

UINT8 is the recommended variant — 75% smaller, 38% faster inference, with virtually no accuracy loss (MAE +0.2ms).

FP16 is not recommended for CPU inference (no native FP16 support, slower than FP32). INT8 (QInt8) is incompatible with some ONNX runtimes due to ConvInteger operator requirements.

Attribution

This is an ONNX format conversion of Meta's MMS forced alignment model, originally distributed via torchaudio.pipelines.MMS_FA.

Original work:

Vineel Pratap, Andros Tjandra, Bowen Shi, et al. "Scaling Speech Technology to 1,000+ Languages." 2023. https://arxiv.org/abs/2305.13516

License

The original model weights are released by Meta under the CC-BY-NC-4.0 license. This ONNX conversion inherits the same license. Non-commercial use only.

Sources

1 source
Hugging FacePrimaryVersion 06e6cbfa5c257d23981c922f6406f5def958eb95
huggingface.co/xycld/lyric-align-mms-fa/tree/06e6cbfa5c257d23981c922f6406f5def958eb95

Verified Sep 13

Model artifacts

2 artifacts

mms_fa_uint8.onnx

onnx · 302 MB · SHA-256 3949a7957375…d42a · Hugging Face

Download

mms_fa.onnx

onnx · 3.10 MB · SHA-256 1de97037f18f…1a47 · Hugging Face

Download

Source excerpts

2 excerpts
ReadmeSep 13huggingface.co/xycld/lyric-align-mms-fa

--- license: cc-by-nc-4.0 language: zh tags: - forced-alignment - ctc - wav2vec2 - onnx - chinese - lyrics - singing pipeline_tag: audio-classification --- # MMS Forced Alignment — ONNX ONNX export of Meta's [MMS_FA (Massively Multilingual Speech Forced Alignment)](https://ai.meta.com/blog/multilingual-model-speech-recognition/) model for CTC forced alignment. ## Files | File | Size | Description | |:-----|:-----|:------------| | `mms_fa.onnx` | 3.2 MB | ONNX model graph | | `mms_fa.onnx.data` | 1.2 GB | External weight data (FP32) | | `tokenizer.json` | 311 B | Token-to-ID mapping (29 tokens) | ## Usage ```python from huggingface_hub import hf_hub_download model_path = hf_hub_download("xycld/lyric-align-mms-fa", "mms_fa.onnx") data_path = hf_hub_download("xycld/lyric-align-mms-fa", "mms_fa.onnx.data") tokenizer_path = hf_hub_download("xycld/lyric-align-mms-fa", "tokenizer.json") ``` **Note:** `mms_fa.onnx.data` must be in the same directory as `mms_fa.onnx` for ONNX Runtime to load correctly. `hf_hub_download` handles this automatically via its cache. ## Model Details - **Original model:** [MMS_FA](https://github.com/facebookresearch/fairseq/tree/main/examples/mms) by Meta Research - **Architecture:** wav2vec2-based CTC forced aligner (315M parameters) - **Precision:** FP32 - **Input:** mono 16kHz waveform - **Output:** log-probability emission matrix [num_frames, 29 labels] at 50fps (20ms/frame) - **ONNX opset:** 18 ## Quantized Variants | Variant | Size | Compression | Load Time | Inference | MAE | Acc @50ms | Acc @100ms | Acc @200ms | Status | |:--------|-----:|:-----------:|----------:|----------:|----:|----------:|-----------:|-----------:|:-------| | **FP32** (original) | 1,207 MB | 1.0x | 989ms | 424ms/line | 34.7ms | 86.2% | 97.5% | 99.4% | Available | | **FP16** | 605 MB | 2.0x | 2,335ms | 576ms/line | 34.7ms | 86.2% | 97.5% | 99.4% | Not recommended | | **UINT8** | 303 MB | 4.0x | 412ms | 262ms/line | 34.9ms | 86.2% | 97.5% | 99.4% | **Recommended** | > Benchmark: Chinese song "错位时空" (362 characters, 53 lines) on CPU. **UINT8 is the recommended variant** — 75% smaller, 38% faster inference, with virtually no accuracy loss (MAE +0.2ms). FP16 is not recommended for CPU inference (no native FP16 support, slower than FP32). INT8 (QInt8) is incompatible with some ONNX runtimes due to `ConvInteger` operator requirements. ## Att...

NoteSep 13huggingface.co/xycld/lyric-align-mms-fa

Source context: 0 downloads · 0 likes · Pipeline audio-classification · Repo xycld/lyric-align-mms-fa