Skip to content

[MAEB] Add proper wav2clip text encoder #3849

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Samoed merged 5 commits into from
Jan 7, 2026
Merged

[MAEB] Add proper wav2clip text encoder #3849

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
576d894
Add proper wav2clip text encoder
AdnanElAssadi Jan 1, 2026
4c9ba89
Add no_revision to revision field
AdnanElAssadi Jan 5, 2026
edfb68c
Added clip revision + torch padding
AdnanElAssadi Jan 5, 2026
02720ab
remove git submodule
AdnanElAssadi Jan 5, 2026
6992a2d
run lint
AdnanElAssadi Jan 6, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions maeb-results
Open in desktop
Submodule maeb-results added at 8ecd55
19 changes: 12 additions & 7 deletions mteb/models/model_implementations/wav2clip_model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,11 +34,16 @@ def __init__(
self.sampling_rate = 16_000
self.max_audio_length_s = max_audio_length_s

# text side (CLIP)
self.clip = CLIPModel.from_pretrained(model_name, revision=revision).to(device)
# text side (CLIP) - we use the standard OpenAI CLIP model as mentioned in paper
# Wav2Clip aligns audio embeddings to CLIP's embedding space using this specific model
clip_model_name = "openai/clip-vit-base-patch32"
clip_revision = "3d74acf9a28c67741b2f4f2ea7635f0aaf6f0268"
self.clip = CLIPModel.from_pretrained(
clip_model_name, revision=clip_revision, device_map=device
)
self.clip.eval()
self.clip_processor = CLIPProcessor.from_pretrained(
model_name, revision=revision
clip_model_name, revision=clip_revision
)

def get_audio_embeddings(
Expand Down Expand Up @@ -71,7 +76,7 @@ def get_audio_embeddings(
orig_freq=sr, new_freq=self.sampling_rate
)
array = resampler(array)
audio_arrays.append(array.numpy())
audio_arrays.append(array) # Keep as tensor

max_length = max(wav.shape[-1] for wav in audio_arrays)
padded_wavs = []
Expand All @@ -84,8 +89,8 @@ def get_audio_embeddings(
padded_wav = wav
padded_wavs.append(padded_wav)

# Stack into batch tensor
batch_tensor = torch.stack(padded_wavs).cpu().detach().numpy()
# Stack into batch array and convert to numpy for embed_audio
batch_tensor = torch.stack(padded_wavs).numpy()

# Process entire batch at once
batch_embeds = self.embed_audio(batch_tensor, self.audio_model)
Expand Down Expand Up @@ -159,7 +164,7 @@ def encode(
loader=Wav2ClipZeroShotWrapper,
name="lyrebird/wav2clip",
languages=["eng-Latn"],
revision="N/A",
revision="no_revision",
release_date="2022-03-15",
modalities=["audio", "text"],
n_parameters=163_000_000, # wav2clip: 11.7M + CLIP: 151.3M ≈ 163M
Expand Down
Loading