yamnet-onnx

Verbatim mirror of zeropointnine/yamnet-onnx, a tf2onnx conversion of Google's official YAMNet (AudioSet, 521 classes).

Re-hosted so it resolves through the uri/rei HF-cache / HF_HUB_OFFLINE=1 path (mic_capture [music_gate.model]).

  • yamnet.onnx โ€” MobileNetV1 backbone, ~3.7 M params, 16 kHz native.
    • input waveform f32 [-1] โ€” raw mono waveform, dynamic length (mel frontend internal)
    • output output_0 f32 [-1, 521] โ€” per-frame AudioSet class scores (Speech=0, Singing=27, Music=132)
    • output output_1 f32 [-1, 1024] โ€” embeddings (unused)
    • output output_2 f32 [-1, 64] โ€” log-mel spectrogram (unused)

License: Apache-2.0 (upstream Google YAMNet).

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support