More work done on the speech system lately. Ben helped with some audio engineering to reduce popping in speech synthesis. Also, we put together a voice for the Bassurians which I really love!
#indiedev #gamedev #indiegame #synth #speechsynth
#indiedev #gamedev #indiegame #synth #speechsynth
March 30, 2026 at 6:19 PM
More work done on the speech system lately. Ben helped with some audio engineering to reduce popping in speech synthesis. Also, we put together a voice for the Bassurians which I really love!
#indiedev #gamedev #indiegame #synth #speechsynth
#indiedev #gamedev #indiegame #synth #speechsynth
January 23, 2026 at 2:04 PM
SwiftF0, a lightweight neural model, achieves state-of-the-art monophonic pitch estimation with 91.80% harmonic mean at 10 dB SNR; trained on speech/music/synthetic data with augmentation; introduces SpeechSynth, a synthetic speech dataset.
SwiftF0: Fast and Accurate Monophonic Pitch Detection
Lars Nieradzik
arxiv.org
August 27, 2025 at 8:23 AM
SwiftF0, a lightweight neural model, achieves state-of-the-art monophonic pitch estimation with 91.80% harmonic mean at 10 dB SNR; trained on speech/music/synthetic data with augmentation; introduces SpeechSynth, a synthetic speech dataset.