Bland Launches Speech v3 Human Speech Engine, Outperforming ElevenLabs and OpenAI
ycombinator · x · 2026-08-05
Bland has introduced Speech v3, billed as the world's first Human Speech Engine. Trained on over 100 million real human conversations, the model targets businesses, developers, and creators.
Speech v3 ranked first in the Design Arena Audio Realism benchmark, beating out major competitors including ElevenLabs, Grok, Cartesia, and OpenAI. To showcase its capabilities, Bland used just 5 seconds of old footage to help a 49-year-old father who lost his voice to a stroke speak again.
Related event: Bland AI Launches Speech v3, Topping Audio Blind Tests(4 posts)→
More from Multimodal
- Photos can be arbitrarily upscaled now, and it's kind of insane — flowersslop · 2026-10-07
- Virtual influencer rendered with Nano Banana 2.1 looks stunning — AIandDesign · 2026-10-07
- 3DGS tracking upgraded: 120fps global-shutter camera doubles pose updates — Scobleizer · 2026-10-07
- FastH3 ships official quantized weights down to 8GB VRAM, Ref2VA reference support coming in V3 — pilkyton · 2026-10-07
- Early ComfyUI Test: Nano Banana 2.1 Shows Better Stability Than 2.0 — Choidonhyeon · 2026-10-07
- ReDetail 2.0: LTX-2.5 Refine-Details LoRA for 4K Upscaling on a 32GB Card — DaLyon92x · 2026-10-07