---
base_model: areffarhadi/Resnet34-tidyvoiceX-ASV
license: apache-2.0
tags:
- speaker-verification
- speaker-embedding
- cross-lingual
- multilingual
- wespeaker
- resnet
- pytorch
datasets:
- voxblink2
- voxceleb2
- tidyvoicex
metrics:
- eer
- mindcf
---

ONNX conversion of [Resnet34-tidyvoiceX-ASV](https://huggingface.co/areffarhadi/Resnet34-tidyvoiceX-ASV) from [TidyVoice Challenge: Cross-Lingual Speaker Verification](https://arxiv.org/abs/2601.21960). Compatitable with ONNX inference script from WeSpeaker and [sherpa-onnx](https://github.com/k2-fsa/sherpa-onnx)

Download
```sh
wget https://huggingface.co/hr16/tidyvoicex-samresnet34-onnx/resolve/main/tidyvoicex_samresnet34.onnx
```

Example on sherpa-onnx
```py
import sherpa_onnx
import soundfile as sf

audio, sample_rate = sf.read("your-audio-file.wav")
embedding_model = sherpa_onnx.SpeakerEmbeddingExtractor(
    sherpa_onnx.SpeakerEmbeddingExtractorConfig(
        model="tidyvoicex_samresnet34.onnx", num_threads=1, provider="cuda"
    )
)
stream = embedding_model.create_stream()
stream.accept_waveform(sample_rate=sample_rate, waveform=audio)
spk_emb = np.array(embedding_model.compute(stream))
```