13 models · ranked by HuggingFace downloads
LAION's CLAP (Contrastive Language-Audio Pretraining) model using the HTSAT (Hierarchical Token-Semantic Audio Transformer) encoder, fused with a text encoder to align audio and text in a shared embedding space. Analogous to CLIP for images, it enables zero-shot audio classification and retrieval using natural language descriptions without task-specific labeled audio data.
wav2vec2-large-robust-24-ft-age-gender performs audio classification by encoding spectral and temporal features to predict one or more discrete labels.
wav2vec2-large-robust-12-ft-emotion-msp-dim maps audio waveforms to class labels. Trained on labeled audio datasets for tasks like language identification and speaker recognition.
wav2vec2-large-xlsr-53-gender-recognition-librispeech is an open-source audio-classification model available on HuggingFace. Details are sourced from the public model registry.
MERT-v1-330M is an open-source audio-classification model available on HuggingFace. Details are sourced from the public model registry.
ast-finetuned-audioset-10-10-0.4593 classifies audio inputs into discrete categories such as language, emotion, speaker identity, or sound event.
emotion-recognition-wav2vec2-IEMOCAP performs audio classification by encoding spectral and temporal features to predict one or more discrete labels.
MuQ-large-msd-iter is an open-source audio-classification model available on HuggingFace. Details are sourced from the public model registry.
An MLX conversion of WeSpeaker's ResNet34 speaker embedding model for Apple Silicon. WeSpeaker-ResNet34 generates d-vector speaker embeddings used for speaker verification and diarization tasks.
open-vakgyata is an open-source audio-classification model available on HuggingFace. Details are sourced from the public model registry.
hubert-large-speech-emotion-recognition-russian-dusha-finetuned is an open-source audio-classification model available on HuggingFace. Details are sourced from the public model registry.
wav2vec-vm-finetune maps audio waveforms to class labels. Trained on labeled audio datasets for tasks like language identification and speaker recognition.
music_genres_classification performs audio classification by encoding spectral and temporal features to predict one or more discrete labels.