H3's voices sound like Microsoft Sam while LTX 2.3 nails tone, user complains
sporkyuncle · reddit · 2026-09-04
A user complains that H3's generated voices sound stiff and bored, like Microsoft Sam — using the bog-standard ComfyUI template with Sage Attention at 20 steps, no amount of descriptive qualifiers changes the flat delivery, especially for male voices.
By contrast, LTX 2.3's voice capabilities are impressive: it picks the right tone on its own and chooses a unique voice per generation (good for variety, bad for reproducibility). The author has repeatedly generated voice with LTX 2.3 and edited it over H3 videos, and asks for advice on improving H3's voices or voice-swap tools.
Related event: H3 Voice Generation Panned as Stiff; LTX 2.3 Outperforms(2 posts)→
More from Multimodal
- Jason Silva Made a 25-Minute Psychedelic AI Film in Five Days by Describing It Out Loud — lmoroney · 2026-10-21
- Ex-Brain Games host made a 25-min AI film in 5 days using 'vibe directing' in iMessage — lmoroney · 2026-10-21
- Motion Launches a General Video Agent That Edits Footage and Mimics Any Reference Style — mhdfaran · 2026-10-21
- Ant open-sources 5 Ming-3.0 speech models spanning TTS, audio understanding and ASR — Med1_Ai · 2026-10-21
- What keeps AI videos watchable: hooks, consistency and a payoff — aftahi_ai · 2026-10-21
- MORPHJUTSU skill in Flova transfers any video's motion to your character — aftahi_ai · 2026-10-21