Datasets:

herwoww
/

arabic_xvector_embeddings

Tasks:

Modalities:

Formats:

Languages:

Size:

Libraries:

License:

Dataset card Data Studio Files Files and versions

herwoww commited on May 13, 2024

Commit

3662f54

·

verified ·

1 Parent(s): 3e49052

Update README.md

Files changed (1) hide show

README.md +1 -1

README.md CHANGED Viewed

@@ -26,7 +26,7 @@ pretty_name: Arabic(M) Speaker Embeddings
 There is one speaker embedding for each utterance in the validation set of both datasets. The speaker embeddings are 512-element X-vectors.
-[Arabic Speech Corpus](https://huggingface.co/datasets/arabic_speech_corpus) has 100 files for a single male speaker and [ClArTTS](https://www.clartts.com/) has 205 files for a single male speaker.
 The X-vectors were extracted using [this script](https://huggingface.co/mechanicalsea/speecht5-vc/blob/main/manifest/utils/prep_cmu_arctic_spkemb.py), which uses the `speechbrain/spkrec-xvect-voxceleb` model.

 There is one speaker embedding for each utterance in the validation set of both datasets. The speaker embeddings are 512-element X-vectors.
+[Arabic Speech Corpus](https://huggingface.co/datasets/arabic_speech_corpus) has 100 files for a single male speaker and [ClArTTS](https://huggingface.co/datasets/MBZUAI/ClArTTS) has 205 files for a single male speaker.
 The X-vectors were extracted using [this script](https://huggingface.co/mechanicalsea/speecht5-vc/blob/main/manifest/utils/prep_cmu_arctic_spkemb.py), which uses the `speechbrain/spkrec-xvect-voxceleb` model.