NV Speech Player DSP Version 6 is here, and it's a big one!

NV Speech Player DSP Version 6 is here, and it's a big one!

The headline feature is the Fujisaki pitch model - a proper prosodic contour system with phrase-level declination and accent commands. Huge thanks to Rommix for the IIR filter implementation and all his work on the pitch code. Natural-sounding prosody is finally possible, from eSpeak's sing-songy style to flatter Eloquence-like delivery.

Coarticulation got a complete overhaul. We've moved to a MITalk-style locus model where vowel START formants shift toward consonant targets, then ramp smoothly to canonical vowel values via new EndCF1-3 parameters. The result is much more natural consonant-to-vowel transitions.

Other highlights:
- Velar pinch for k/g fixes confusion between words like "script" and "stripped"
- Hungarian palatal stops (gy/ty) now sound correct with proper formants and j off-glides
- Hungarian vowels tuned based on phonetic research
- British English PALM/START vowel no longer sounds Australian
- Breathiness at 100% actually sounds breathy now (yes, really)

This release maintains backward compatibility - new features are opt-in via YAML config.
Enjoy - I'm sure the Fujisaki defaults will get way more tuned as people provide feedback on what worked for them to improve it. New settings are in the language panel!
https://github.com/tgeczy/NVSpeechPlayer/releases/download/v-265/nvSpeechPlayer-2026-v265.nvda-addon
https://github.com/tgeczy/NVSpeechPlayer/releases/download/v-265/nvSpeechPlayer-linux-x86_64-v265.tar.gz
https://github.com/tgeczy/NVSpeechPlayer/releases/download/v-265/NVSPPhonemeEditor-v265.zip