MOS TTS multilanguage reference phrases: Difference between revisions
Jump to navigation
Jump to search
(Initial version of reference phrases for MOS comparison) |
No edit summary |
||
| Line 1: | Line 1: | ||
==MOS (Mean Opinion Score)== | [[Category:TTS models]] | ||
== MOS (Mean Opinion Score) == | |||
The MOS (Mean Opinion Score) is a numeric value to define a voice quality between 1 (''bad'') and 5 (''good''). To make a reliable MOS comparison we should define default phrases (for several languages) for all TTS models. This should avoid "cherry picking" to generate best results based on a specific dataset and TTS model. | The MOS (Mean Opinion Score) is a numeric value to define a voice quality between 1 (''bad'') and 5 (''good''). To make a reliable MOS comparison we should define default phrases (for several languages) for all TTS models. This should avoid "cherry picking" to generate best results based on a specific dataset and TTS model. | ||
==Reference phrases for MOS comparion== | == Reference phrases for MOS comparion == | ||
===English=== | === English === | ||
* "It took me quite a long time to develop a voice, and now that I have it I'm not going to be silent." | * "It took me quite a long time to develop a voice, and now that I have it I'm not going to be silent." | ||
* "Be a voice, not an echo." | * "Be a voice, not an echo." | ||
| Line 10: | Line 12: | ||
* ''feel free to add items here'' | * ''feel free to add items here'' | ||
===German=== | === German === | ||
* "Es dauert lange, eine eigene Stimme zu entwickeln, aber jetzt wo ich sie habe, bin ich nie wieder still." | * "Es dauert lange, eine eigene Stimme zu entwickeln, aber jetzt wo ich sie habe, bin ich nie wieder still." | ||
* ''feel free to add items here'' | * ''feel free to add items here'' | ||
=== French === | |||
=== Spanish === | |||
Latest revision as of 23:39, 3 November 2021
MOS (Mean Opinion Score)[edit | edit source]
The MOS (Mean Opinion Score) is a numeric value to define a voice quality between 1 (bad) and 5 (good). To make a reliable MOS comparison we should define default phrases (for several languages) for all TTS models. This should avoid "cherry picking" to generate best results based on a specific dataset and TTS model.
Reference phrases for MOS comparion[edit | edit source]
English[edit | edit source]
- "It took me quite a long time to develop a voice, and now that I have it I'm not going to be silent."
- "Be a voice, not an echo."
- "I'm sorry Dave. I'm afraid I can't do that."
- "This cake is great. It's so delicious and moist."
- feel free to add items here
German[edit | edit source]
- "Es dauert lange, eine eigene Stimme zu entwickeln, aber jetzt wo ich sie habe, bin ich nie wieder still."
- feel free to add items here