Chimpanzee Vocalization CNN10 (sanctuary data only)

A 10-layer Convolutional Neural Network for detecting chimpanzee vocalizations in bioacoustic recordings. Part of the animal-sounds pipeline developed at Utrecht University.

Model description

• Architecture: CNN10 — 4 convolutional blocks (64→128→256→512 channels), global average pooling, fully-connected head, 2-class output (chimpanzee vocalization vs. background)
• Input: 3-channel mel spectrogram representation (mel spectrogram → PCEN normalization → per-file z-normalization → delta features)
• Preprocessing: recordings resampled to 48,000 Hz, filtered with a Butterworth bandpass filter (100–2,000 Hz)
• Frame length: 0.5-second audio frames

This checkpoint was trained without synthetic data — real chimpanzee vocalizations and background recordings from Cameroon only.

A version trained with synthetic augmentation for improved cross-environment generalization is also available: utrechtuniversity/chimp-vocalization-cnn10-synthetic — recommended as the default choice for most use cases involving recordings from environments different from the training sanctuary.

Training data

• Sanctuary recordings only: Mefou Primate Sanctuary, Cameroon (December 2019 – January 2020), publicly available (Zwerts et al., 2024)

Evaluation

Evaluated at file level on held-out recordings from Chimfunshi Wildlife Orphanage Trust (CWOT), Zambia — a different acoustic environment from the training data, used to assess cross-environment generalization. Two recorders ("13b" and "14a") were evaluated separately.

Metrics are for the positive (chimpanzee) class unless labeled "-macro" (averaged across both classes).

Note this model tends to show higher precision but lower recall on the cross-environment (Zambia) test set compared to the synthetic-augmented variant, reflecting that it was not exposed to Zambian background acoustics during training.

Usage

This model is not integrated with transformers — it's a plain PyTorch model exported with PyTorchModelHubMixin. Copy the class definition below, then load the trained weights directly from the Hub.

import torch.nn as nn
from huggingface_hub import PyTorchModelHubMixin


def _cnn_block(in_ch, out_ch):
    return nn.Sequential(
        nn.Conv2d(in_ch, out_ch, kernel_size=3, padding=1),
        nn.BatchNorm2d(out_ch),
        nn.ReLU(inplace=True),
    )


class CNN10Hub(nn.Module, PyTorchModelHubMixin):
    def __init__(self, num_channels=3, num_labels=2, dropout_rate=0.5, **kwargs):
        super().__init__()
        layers = [
            _cnn_block(num_channels, 64), _cnn_block(64, 64), nn.AvgPool2d(2), nn.Dropout(dropout_rate),
            _cnn_block(64, 128), _cnn_block(128, 128), nn.AvgPool2d(2), nn.Dropout(dropout_rate),
            _cnn_block(128, 256), _cnn_block(256, 256), nn.AvgPool2d(2), nn.Dropout(dropout_rate),
            _cnn_block(256, 512), _cnn_block(512, 512), nn.AvgPool2d(2), nn.Dropout(dropout_rate),
            nn.AdaptiveAvgPool2d(1), nn.Dropout(dropout_rate), nn.Flatten(),
            nn.Linear(512, 512), nn.ReLU(), nn.Dropout(dropout_rate),
            nn.Linear(512, num_labels),
        ]
        self.acoustic_model = nn.Sequential(*layers)

    def forward(self, x):
        return self.acoustic_model(x)


model = CNN10Hub.from_pretrained("utrechtuniversity/chimp-vocalization-cnn10-sanctuary")
model.eval()

Important: the model expects input tensors matching the exact preprocessing pipeline described above (3-channel mel + PCEN + delta features, 48kHz source audio, 0.5s frames). Raw audio or standard mel spectrograms without this preprocessing will produce incorrect predictions.

A standalone preprocess.py (dependent only on librosa, numpy, scipy — no need to install the full source package) is included in this repo, reproducing the exact training-time feature extraction:

from huggingface_hub import hf_hub_download
import importlib.util
import torch

path = hf_hub_download("utrechtuniversity/chimp-vocalization-cnn10-sanctuary", "preprocess.py")
spec = importlib.util.spec_from_file_location("preprocess", path)
preprocess = importlib.util.module_from_spec(spec)
spec.loader.exec_module(preprocess)

features = preprocess.extract_features("my_recording.wav")   # (3, n_mel, n_frames)
chunks = preprocess.chunk_features(features)                 # (n_chunks, 3, n_mel, 64)

with torch.no_grad():
    probs = torch.softmax(model(torch.from_numpy(chunks).float()), dim=1)

See the feature extraction source in the GitHub repository for further background on the pipeline.

Intended use and limitations

• Trained specifically for chimpanzee vocalization detection at the sanctuary recording conditions described above. Cross-environment performance (e.g. on the Zambia test set above) is lower than the synthetic-augmented variant, particularly in recall.
• Useful as a baseline for comparing the effect of synthetic cross-environment augmentation, or in settings closely matching the original Cameroon recording conditions.
• Although developed for chimpanzee vocalizations, the underlying pipeline in the source repository can be adapted to other species given labeled training data.

License

Code released under Apache 2.0, matching the source GitHub repository.

Citation

The chimpanzee vocalization dataset used to train this model:

Zwerts, J. A., Treep, J., Kaandorp, C. S., Meewis, F., Koot, A. C., & Kaya, H. (2021).
Introducing a central african primate vocalisation dataset for automated species classification.
arXiv preprint. https://arxiv.org/pdf/2101.10390.pdf

Zwerts, J., Treep, J., Zahedi, P., & Kaandorp, C. (2024).
Central African Primate vocalization bioacoustics dataset: Yoda Data publication platform of Utrecht University.

A paper describing the CNN10/CNN12 models and cross-environment evaluation presented here is currently in preparation. This model card will be updated with the correct citation once available — in the meantime, please contact Dr. Joeri A. Zwerts (j.a.zwerts@uu.nl) for citation guidance.

Links

• Source code and full pipeline: github.com/UtrechtUniversity/animal-sounds
• Contact: Dr. Joeri A. Zwerts, Assistant Professor — j.a.zwerts@uu.nl
• Technical/repository support: Research Engineering team, Utrecht University — research.engineering@uu.nl