Spaces:
Running on Zero
Running on Zero
Card shows both voices
Browse files- README.md +4 -0
- samples/voices.jpg +0 -0
README.md
CHANGED
|
@@ -12,6 +12,10 @@ short_description: A 4.5M parameter two-voice English TTS, instant
|
|
| 12 |
|
| 13 |
# tiny-voca text to speech
|
| 14 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
A text to speech model with 4.5 million parameters, trained from scratch on about 18 hours
|
| 16 |
of synthetic speech. Two voices, F2 and M2, 24 kHz mono. It runs roughly 70 times faster than
|
| 17 |
real time on four CPU threads, so a sentence comes back in well under a second.
|
|
|
|
| 12 |
|
| 13 |
# tiny-voca text to speech
|
| 14 |
|
| 15 |
+

|
| 16 |
+
|
| 17 |
+
The same line through both voices, same seed. F2 takes 2.41 s, M2 takes 2.09 s, and the shape of the phrase differs as much as the pitch does.
|
| 18 |
+
|
| 19 |
A text to speech model with 4.5 million parameters, trained from scratch on about 18 hours
|
| 20 |
of synthetic speech. Two voices, F2 and M2, 24 kHz mono. It runs roughly 70 times faster than
|
| 21 |
real time on four CPU threads, so a sentence comes back in well under a second.
|
samples/voices.jpg
ADDED
|