Spestly/Axis-1-14B
Updated
Aayan Mishra
Spestly
AI & ML interests
None yet
Recent Activity
published
a model
2 days ago
Spestly/Axis-1-14B
updated
a model
5 days ago
Spestly/Everest-N1-360M
published
a model
5 days ago
Spestly/Everest-N1-360M
Organizations
Spestly's activity
published
a
model
2 days ago
updated
a
model
5 days ago
published
a
model
5 days ago
published
a
model
9 days ago
updated
a
collection
11 days ago
Error in benchmarking model - rubenroy/Gilgamesh-72B
2
#1115 opened 12 days ago
by
Spestly
liked
a
model
12 days ago
liked
a
Space
13 days ago
updated
a
model
13 days ago
published
a
model
15 days ago
reacted to
hexgrad's
post with 🔥
15 days ago
Post
5640
I wrote an article about G2P: https://hf.co/blog/hexgrad/g2p
G2P is an underrated piece of small TTS models, like offensive linemen who do a bunch of work and get no credit.
Instead of relying on explicit G2P, larger speech models implicitly learn this task by eating many thousands of hours of audio data. They often use a 500M+ parameter LLM at the front to predict latent audio tokens over a learned codebook, then decode these tokens into audio.
Kokoro instead relies on G2P preprocessing, is 82M parameters, and thus needs less audio to learn. Because of this, we can cherrypick high fidelity audio for training data, and deliver solid speech for those voices. In turn, this excellent audio quality & lack of background noise helps explain why Kokoro is very competitive in single-voice TTS Arenas.
G2P is an underrated piece of small TTS models, like offensive linemen who do a bunch of work and get no credit.
Instead of relying on explicit G2P, larger speech models implicitly learn this task by eating many thousands of hours of audio data. They often use a 500M+ parameter LLM at the front to predict latent audio tokens over a learned codebook, then decode these tokens into audio.
Kokoro instead relies on G2P preprocessing, is 82M parameters, and thus needs less audio to learn. Because of this, we can cherrypick high fidelity audio for training data, and deliver solid speech for those voices. In turn, this excellent audio quality & lack of background noise helps explain why Kokoro is very competitive in single-voice TTS Arenas.
upvoted
an
article
15 days ago
Article
G2P Shrinks Speech Models
By
•
•
35
upvoted
a
collection
17 days ago
updated
a
Space
18 days ago
liked
a
model
19 days ago
updated
a
model
20 days ago
published
a
model
20 days ago
Datasets and Training Process
2
#6 opened 21 days ago
by
fhai50032
updated
a
model
21 days ago