Is it possible to export the model.encode_code part to ONNX? I tried but encountered an error during export. I want to use this tokenizer for feature extraction.

[whaozl]

Is it possible to export the model.encode_code part to ONNX? I tried but encountered an error during export. I want to use this tokenizer for feature extraction.

import librosa
import torch
import torchaudio
from torchaudio import transforms as T
from neucodec import NeuCodec
 
model = NeuCodec.from_pretrained("neuphonic/neucodec")
model.eval().cuda()   
 
y, sr = torchaudio.load(librosa.ex("libri1"))
if sr != 16_000:
    y = T.Resample(sr, 16_000)(y)[None, ...] # (B, 1, T_16)

with torch.no_grad():
    fsq_codes = model.encode_code(y)

[KevinAHM]

https://huggingface.co/KevinAHM/distill-neucodec-onnx

[whaozl]

Hello, is there a way to export the encoder of the original model https://huggingface.co/neuphonic/neucodec, rather than the distilled version?

Requesting a help, thank you.

@KevinAHM