This model is a fine-tuned version of facebook/wav2vec2-base-960h for Speech Emotion Recognition (SER). It has been trained using a Frozen Feature Extractor strategy to preserve the model's acoustic understanding while adapting to emotion detection. This approach ensures stable performance and prevents "Catastrophic Forgetting," achieving nearly 80% accuracy on the validation set. Update: The "Calm" and "Neutral" classes have been merged to improve classification consistency, resulting in 7 distinct emotion classes. The model was trained on a combined dataset of ~12,000 audio files from: The model classifies audio into one of the following emotions: 1. Angry 2. Disgust 3. Fear 4. Happy 5.…
Open-weight model · Audio classification
wav2vec2-base-drum-kit
by Andrew Keig airasoul/wav2vec2-base-drum-kit
Fine-tuned facebook/wav2vec2-base for audio classification of single drum/percussion sounds into 10 classes. - clap, conga, crash, cymbal, hat, kick, ride, rim, snare, tom - Trained on short, single-hit drum sounds.
Parameters95M
Context—
Weights378.3 MB
Licensemit
AccessOpen weights
Monthly Downloads8.5k
Runs On
What it takes to serve wav2vec2-base-drum-kit (95M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 0.2 GB | 0.2 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.1 GB | 0.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.0 GB | 0.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Sep 18, 2026.
Model Card
By Andrew Keig, published under mit, revision 1d2aca50b37c.
Fine-tuned facebook/wav2vec2-base for audio classification of single drum/percussion sounds into 10 classes. - clap, conga, crash, cymbal, hat, kick, ride, rim, snare, tom - Trained on short, single-hit drum sounds. Performance may drop on long mixes, multiple overlapping sounds, or very different recording conditions.
Read Andrew Keig's full model card
Wav2Vec2 for drum-kit classification
Fine-tuned facebook/wav2vec2-base for audio classification of single drum/percussion sounds into 10 classes.
Classes
- clap, conga, crash, cymbal, hat, kick, ride, rim, snare, tom
Usage
import torch
import librosa
from transformers import AutoFeatureExtractor, AutoModelForAudioClassification
model_id = "airasoul/wav2vec2-base-drum-kit" # e.g. username/wav2vec2-base-drum-kit
feature_extractor = AutoFeatureExtractor.from_pretrained(model_id)
model = AutoModelForAudioClassification.from_pretrained(model_id)
model.eval()
# Load a WAV (16 kHz mono)
audio, sr = librosa.load("path/to/audio.wav", sr=16000, mono=True)
inputs = feature_extractor(
audio, sampling_rate=16000, max_length=48000, # 3 s at 16 kHz
truncation=True, return_tensors="pt", padding=True
)
with torch.no_grad():
logits = model(**inputs).logits
pred_id = logits.argmax(dim=-1).item()
label = model.config.id2label.get(pred_id) or model.config.id2label.get(str(pred_id))
print(label) # e.g. "kick"
Training
- Base: facebook/wav2vec2-base
- Task: Single-label classification over 10 drum classes
- Data: Custom drum-kit dataset with augmentation (time stretch, noise, gain)
- Input: 16 kHz mono, up to 3 s (truncated or padded)
Results
- Validation accuracy: 95.7% (epoch 10)
- Test accuracy: 97.0% (300 samples, held-out)
| Class | Precision | Recall | F1-score |
|---|---|---|---|
| clap | 1.00 | 1.00 | 1.00 |
| conga | 0.96 | 0.93 | 0.95 |
| crash | 0.97 | 0.97 | 0.97 |
| cymbal | 1.00 | 0.91 | 0.95 |
| hat | 1.00 | 0.97 | 0.98 |
| kick | 1.00 | 0.94 | 0.97 |
| ride | 0.94 | 1.00 | 0.97 |
| rim | 1.00 | 1.00 | 1.00 |
| snare | 0.89 | 0.96 | 0.93 |
| tom | 0.92 | 1.00 | 0.96 |
Limitations
- Trained on short, single-hit drum sounds. Performance may drop on long mixes, multiple overlapping sounds, or very different recording conditions.
Configuration
- Architecture
- Wav2Vec2ForSequenceClassification
- Layers
- 12
- Hidden size
- 768
- Feed-forward size
- 3,072
- Attention heads
- 12
- Vocabulary size
- 32
- Model type
- wav2vec2
Identity and Version
- Repository
- airasoul/wav2vec2-base-drum-kit
- Publisher
- Andrew Keig
- Task
- Audio classification
- Modality
- Audio
- Library
- Not stated by the source
- Parameters
- 95M parameters
- Languages
- en
- Revision
- 1d2aca50b37ca14af53f6a7b79608255f8a46d13
- First published
- 2026-02-10
- Last updated
- 2026-02-10
Files and Weights
5 files, 378.3 MB in total. The weights are 1 file totalling 378.3 MB in safetensors.
Weights1 file · 378.3 MB
Configuration2 files · 2.9 KB
Documentation1 file · 2.4 KB
Repository1 file · 1.5 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 378.3 MB | 2eb6f79aa010 |
| config.json | Configuration | 2.7 KB | — |
| preprocessor_config.json | Configuration | 215 B | — |
| README.md | Documentation | 2.4 KB | — |
| .gitattributes | Repository | 1.5 KB | — |
License and Download
- License
- mit
- Access
- Open weights, no gate
- Download size
- 378.3 MB
Download from Andrew Keig
Released by Andrew Keig through its official repository on Hugging Face. Read the license.
Built From
- Trained on (disclosed) airasoul/drum-kit
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 378.3 MB |
| 16-bit | 0.2 GB |
| 8-bit | 0.1 GB |
| 4-bit | 0.0 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About wav2vec2-base-drum-kit
How much GPU memory does wav2vec2-base-drum-kit need?
About 0.2 GB at 16-bit and 0.1 GB at 4-bit: the weights (95M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run wav2vec2-base-drum-kit on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use wav2vec2-base-drum-kit commercially?
Yes. wav2vec2-base-drum-kit is released under MIT License. The MIT License is a short permissive license. It permits commercial use, modification and redistribution, provided the copyright notice and permission notice are included.
Similar Models
Wav2Vec2: Self-Supervised Learning for Speech Recognition: https://arxiv.org/pdf/2006.11477 male female Common-Voice-Gender-Detection is designed for: Speech Analytics – Assist in analyzing speaker demographics in call centers or customer service recordings. Conversational AI Personalization – Adjust tone or dialogue based on gender detection for more personalized voice assistants. Voice Dataset Curation – Automatically tag or filter voice datasets by speaker gender for better dataset management. Research Applications – Enable linguistic and acoustic research involving gender-specific speech patterns. Multimedia Content Tagging – Automate metadata generation for gender identification in…
This model is a fine-tuned version of mo-thecreator/wav2vec2-base-finetuned on the None dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 3e-05 - trainbatchsize: 8 - evalbatchsize: 8 - gradientaccumulationsteps: 4 - totaltrainbatchsize: 32 - lrschedulertype: linear - lrschedulerwarmupratio: 0.1 - numepochs: 5 - Transformers 4.39.3 - Pytorch 2.1.2 - Datasets 2.18.0 - Tokenizers 0.15.2 - mo-thecreator
This model is a fine-tuned version of motheecreator/Deepfake-audio-detection on the audiofolder dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 3e-05 - trainbatchsize: 32 - evalbatchsize: 32 - gradientaccumulationsteps: 4 - totaltrainbatchsize: 128 - lrschedulertype: cosine - lrschedulerwarmupratio: 0.1 - numepochs: 5 - Transformers 4.41.2 - Pytorch 2.1.2 - Datasets 2.19.2 - Tokenizers 0.19.1
The model expects a raw audio signal as input and outputs predictions for age in a range of approximately 0...1 (0...100 years) and gender expressing the probababilty for being child, female, or male. In addition, it also provides the pooled states of the last transformer layer. The model was created by fine-tuning Wav2Vec2-Large-Robust Timit and For this version of the model we only trained the first six transformer layers. An ONNX export of the model is available from Further details are given in the associated paper and tutorial.
Model · Audio classification
ast-finetuned-audioset-10-10-0.4593
Audio Spectrogram Transformer (AST) model fine-tuned on AudioSet. It was introduced in the paper AST: Audio Spectrogram Transformer by Gong et al. and first released in this repository. Disclaimer: The team releasing Audio Spectrogram Transformer did not write a model card for this model so this model card has been written by the Hugging Face team. The Audio Spectrogram Transformer is equivalent to ViT, but applied on audio. Audio is first turned into an image (as a spectrogram), after which a Vision Transformer is applied. The model gets state-of-the-art results on several audio classification benchmarks. You can use the raw model for classifying audio into one of the AudioSet classes. See…