ConvFSENet

A causal, fully-convolutional speech enhancer trained on VoiceBank-DEMAND-16k. Source: github.com/LarocheC/sparse-nsnet2. See RESULTS_CONVFSENET.md for the full results, architecture description, and the magnitude-compression trick that makes int8 deployment essentially loss-free.

Headline numbers

metric	value
params	1.45 M
FP32 PESQ	2.931
int8 PESQ	2.911
Δ (FP32→int8)	+0.020
int8 RTF (ORT CPU)	0.017
int8 size	1.6 MiB

PESQ is on the full 824-utterance VoiceBank-DEMAND test split.

Files

file	what it is
g_best	PyTorch checkpoint (full state dict — generator, optim, etc.)
g_best_fp32.onnx	Streaming FP32 ONNX (per-frame inputs + FIFO state buffers)
g_best.onnx	Static int8 ONNX (QDQ, per-channel weights, MinMax calibration; compression prologue kept FP32)
config.json	Training config (architecture + STFT params)

Loading

PyTorch:

import json, torch
from huggingface_hub import hf_hub_download
from common.env import AttrDict
from convfsenet.model import build_causal_model

REPO = "claroche1/convfsenet"
cfg  = json.load(open(hf_hub_download(REPO, "config.json")))
ckpt = torch.load(hf_hub_download(REPO, "g_best"),
                  map_location="cuda", weights_only=False)
model = build_causal_model(AttrDict(cfg)).cuda().eval()
model.load_state_dict(ckpt["generator"])

ONNX (FP32 or int8):

import onnxruntime as ort
from huggingface_hub import hf_hub_download

REPO = "claroche1/convfsenet"
sess = ort.InferenceSession(
    hf_hub_download(REPO, "g_best.onnx"),       # or g_best_fp32.onnx
    providers=["CPUExecutionProvider"],
)
# Streaming shape: feed one frame of magnitude STFT (B, n_freq) + the per-block
# FIFO state buffers per call. End-to-end RMS-norm + STFT + frame loop + iSTFT
# pipeline lives in convfsenet/inference_onnx.py in the source repo.

License

MIT. See the source repository for training code and full attribution.