oscarfree

5 ms·

1. 1.▲
   by oscarfree 11mo ago

   We are working on this - we don't have quite enough Irish speech data.
2. 2.▲
   by oscarfree 11mo ago

   Part of this is the "dark L" sound
3. 3.▲
   by oscarfree 11mo ago

   12 layers of 768-dim each. The 3 dimensions visualized are chosen by UMAP.
4. 4.▲
   by oscarfree 11mo ago

   Not sure, could be the large number of Spanish dialects represented in the dataset, label noise, or something else. There may just be too much diversity in the class to fit neatly in a cluster. Also, the training dataset is highly imbalance
5. 5.▲
   by oscarfree 11mo ago

   We didn't explicitly. Because we finetuned this model for accent classification, the later transformer layers appear to ignore non-accent vocal characteristics. I verified this for gender for example.
6. 6.▲
   by oscarfree 1y ago

   People have all sorts of motivations for learning languages and accents. Right now, I'm using this tech to work on my accent in Spanish. Honestly I would rather mumble almost unintelligibly with an decent Mexican accent than speak Span