Testes de desempenho de mídia
Compare a qualidade de geração entre modelos de mídia
Por que escolhemos este prompt: Parenthetical cues like (sighs) and (laughs) ask for a sound, not a word. This checks whether the model produces the nonverbal, skips it, or reads the cue text out loud.
(sighs) Route 6 Espresso is open till nine. (laughs) The last train leaves at eleven fifty-six.
padrão do provedor
george
george
george
george
padrão do provedor
padrão do provedor
padrão do provedor
Zephyr
Zephyr
English_expressive_narrator

longanlingxin
eve
Zephyr

cs-CZ-Grant:MAI-Voice-2.1-Flash
English_expressive_narrator

loongjohn

cs-CZ-Grant:MAI-Voice-2.1
flux-alexis-en
padrão do provedor
george
george
george
george
george

af_alloy
nadine
nadine

en-US-Harper:MAI-Voice-2

en-US-Harper:MAI-Voice-2

en_paul_sad
conversational_a
padrão do provedor
aura-2-thalia-en
tara

voice_3150