This model is a fine-tuned version of Kkonjeong/wav2vec2-base-korean on the None dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 0.0001 - trainbatchsize: 8 - evalbatchsize: 8 - lrschedulertype: linear - numepochs: 10 - mixedprecisiontraining: Native AMP - Transformers 5.18.0 - Pytorch 2.11.0+cpu - Datasets 4.8.5 - Tokenizers 0.23.2
SAVRN Model Hub · Models by Task
Audio Classification Models
130 open-weight audio classification models in the SAVRN Model Hub, with Quan, AI at Meta and Superb publishing the most.
130Models
81Publishers
309K to 1.5BParameter range
8Licenses
SAVRN's Take
Audio classification covers everything from tagging the sounds in a clip to judging a speaker's emotion, age or gender. The most downloaded model here, LAION's CLAP, works like CLIP for sound: describe a category in words and it scores audio against it, with no training. That flexibility is why it leads by a wide margin.
Most of these models are small enough to run on a CPU, so hardware is rarely the constraint. Licensing and purpose are. A meaningful share of this category carries noncommercial terms, including audEERING's age, gender and emotion models, and inferring personal traits from a voice raises privacy questions in many jurisdictions whatever the license says. For commercial sound tagging, the Apache and BSD options, CLAP and MIT's Audio Spectrogram Transformer among them, are the clean starting points.
Most Downloaded
| Model | Publisher | Parameters | License | Monthly downloads | Cheapest GPUs at 16-bit |
|---|---|---|---|---|---|
| clap-htsat-fused | LAION eV | 154M | apache-2.0 | 7.4M | 1x MI300X, $1.85/hr |
| wav2vec2-large-robust-24-ft-age-gender | audEERING GmbH | 318M | cc-by-nc-sa-4.0 | 2.3M | 1x MI300X, $1.85/hr |
| ast-finetuned-audioset-10-10-0.4593 | Massachusetts Institute of Technology | 87M | bsd-3-clause | 1.3M | 1x MI300X, $1.85/hr |
| ced-gguf | Mudler | — | apache-2.0 | 1M | — |
| wav2vec2-large-robust-12-ft-emotion-msp-dim | audEERING GmbH | 165M | cc-by-nc-sa-4.0 | 804.8k | 1x MI300X, $1.85/hr |
| wav2vec2-large-xlsr-53-gender-recognition-librispeech | Alef Iury Siqueira Ferreira | 316M | apache-2.0 | 780.7k | 1x MI300X, $1.85/hr |
| audiobox-aesthetics | AI at Meta | 104M | cc-by-4.0 | 635.2k | 1x MI300X, $1.85/hr |
| open-vakgyata | OneCXi | 59M | cc-by-nc-4.0 | 500.7k | 1x MI300X, $1.85/hr |
| MuQ-large-msd-iter | MuQ | 333M | cc-by-nc-4.0 | 328.2k | 1x MI300X, $1.85/hr |
| hubert-large-speech-emotion-recognition-russian-dusha-finetuned | Phil | 316M | apache-2.0 | 218.3k | 1x MI300X, $1.85/hr |
Licenses
| License | Models | Commercial use |
|---|---|---|
| apache-2.0 | 48 | Yes |
| not stated | 46 | Not stated |
| mit | 11 | Yes |
| cc-by-nc-4.0 | 11 | Not without separate permission |
| cc-by-nc-sa-4.0 | 4 | Not without separate permission |
| cc-by-4.0 | 4 | Yes |
| bsd-3-clause | 4 | Yes |
| openrail | 2 | Yes, with conditions |
Who Publishes Them
All 130 Models, Page 3 of 3
This model is a fine-tuned version of Kkonjeong/wav2vec2-base-korean on the None dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 0.0001 - trainbatchsize: 8 - evalbatchsize: 8 - lrschedulertype: linear - numepochs: 10 - mixedprecisiontraining: Native AMP - Transformers 5.18.0 - Pytorch 2.11.0+cu130 - Datasets 4.8.5 - Tokenizers 0.23.2
This model is a fine-tuned version of Kkonjeong/wav2vec2-base-korean on the None dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 0.0001 - trainbatchsize: 8 - evalbatchsize: 8 - lrschedulertype: linear - numepochs: 10 - mixedprecisiontraining: Native AMP - Transformers 5.18.0 - Pytorch 2.11.0+cpu - Datasets 4.8.5 - Tokenizers 0.23.2
This model is a fine-tuned version of Kkonjeong/wav2vec2-base-korean on the None dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 0.0001 - trainbatchsize: 8 - evalbatchsize: 8 - lrschedulertype: linear - numepochs: 10 - mixedprecisiontraining: Native AMP - Transformers 5.18.0 - Pytorch 2.11.0+cpu - Datasets 4.8.5 - Tokenizers 0.23.2
This model is a fine-tuned version of Kkonjeong/wav2vec2-base-korean on the None dataset. It achieves the following results on the evaluation set: - evalloss: 0.5162 - evalaccuracy: 1.0 - evalruntime: 20.8464 - evalsamplespersecond: 0.911 - evalstepspersecond: 0.144 The following hyperparameters were used during training: - learningrate: 0.0001 - trainbatchsize: 8 - evalbatchsize: 8 - lrschedulertype: linear - numepochs: 10 - mixedprecisiontraining: Native AMP - Transformers 5.18.0 - Pytorch 2.11.0+cpu - Datasets 4.8.5 - Tokenizers 0.23.2
This model is a fine-tuned version of Kkonjeong/wav2vec2-base-korean on the None dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 0.0001 - trainbatchsize: 8 - evalbatchsize: 8 - lrschedulertype: linear - numepochs: 10 - mixedprecisiontraining: Native AMP - Transformers 5.18.0 - Pytorch 2.11.0+cu130 - Datasets 4.8.5 - Tokenizers 0.23.2
This model is a fine-tuned version of Kkonjeong/wav2vec2-base-korean on the None dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 0.0001 - trainbatchsize: 8 - evalbatchsize: 8 - lrschedulertype: linear - numepochs: 10 - mixedprecisiontraining: Native AMP - Transformers 5.18.0 - Pytorch 2.11.0+cu130 - Datasets 4.8.5 - Tokenizers 0.23.2
This model is a fine-tuned version of Kkonjeong/wav2vec2-base-korean on the None dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 0.0001 - trainbatchsize: 4 - evalbatchsize: 4 - lrschedulertype: linear - numepochs: 10 - mixedprecisiontraining: Native AMP - Transformers 5.18.0 - Pytorch 2.11.0+cpu - Datasets 4.8.5 - Tokenizers 0.23.2
This model is a fine-tuned version of Kkonjeong/wav2vec2-base-korean on the None dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 0.0001 - trainbatchsize: 4 - evalbatchsize: 4 - lrschedulertype: linear - numepochs: 10 - mixedprecisiontraining: Native AMP - Transformers 5.18.0 - Pytorch 2.11.0+cpu - Datasets 4.8.5 - Tokenizers 0.23.2
Ready wake-word models for OpenVoiceOS, trained with wakeforge. Each model is a small GRU classifier that reads features from WakeHuBERT tiny, a 0.64M-parameter speech feature extractor distilled from HuBERT-base. A model scores the last 1.5 s of audio (75 feature frames) and outputs a logit; its sigmoid is the probability that the window holds the wake word. Try them in your browser: WakeHuBERT wake-words Space runs every model here on your microphone or an uploaded file, with a live score and the calibrated threshold. Everything runs locally in the browser. The repository holds one ONNX file per word under models/, and models.json, which lists each model's word, featurizer, calibration…
Questions
Which Audio classification models are most downloaded?
By monthly downloads reported by the Hugging Face Hub: .