DRAFT
Soudscapes via AudioGen
|
|
|
|---|---|
|
See level classhing on the iceberg of lighhouse |
|
|
Loug scene of restaurant in medieval city |
|
|
Nikolskoe bei der Wasser an der Havel in Berlin-Zehlendorf, nahe der Pfaueninsel. Beliebtes Berliner Ausflugsziel |
Draft after this line
Following examples are the Harvard sentences synthesized via StyleTTS2 - using Mimic-3 or accelerated 4x speed Mimic-3 styles or Librispeech segments
Trial 3
New Tablo - tts_harvard.py =======================================================
|
Prompt - ( |
StyleTTS2 - (First 20 Harvard Sentences) |
|---|---|
|
Mimic-3 English |
StyleTTS - (Mimic-3 English) From Above |
|
Mimic-3 English 4x |
StyleTTS2 - (Mimic-3 English 4x) |
|
Human |
StyleTTS2 - (Human) |
|
Mimic-3 Foreign |
StyleTTS2 - (Mimic-3 Foreign) |
|
Mimix-3 Foreign 4x |
StyleTTS2 - (Mimic-3 Foreign 4x) |
Please edit for MOS annotation
|
|
|
|
0 |
1 |
|
2 |
3 |
|
4 5 |
6 7 |
|
8 |
9 |
|
10 |
11 |
4 4
1 4
5 5
5 4
3 4
1 4
2 3
1 4
5 4
5 4
2 3
1 2
A B
4* 5
3* 5
5 5
5 4
2 5
1 4
*ignoring the breath-noise between sentences