Klatt formant synthesis voicing formats demonstration

March 1, 2026·14 min·1 clip

A formant synthesis demo shows how compression formats destroy or fuzz waveforms at different sampling rates.

The episode is a solo demonstration by the host experimenting with formant synthesis and audio compression formats. It begins with an introduction to the setup, using a formant synthesis utterance and two sampling rates: 8,000 Hz for telephone quality and 16,000 Hz for neuroband. The host explains the goal is to examine how different formats either destroy or preserve waveforms, making them sound fuzzy or extra vocalized. The first example is GSM compression, which makes the audio sound grainy compared to the original utterance. Next, the host steps through AMR compression at various bit rates from 4.75 to 12.2 kbps, noting that lower bit rates result in fuzzier and more compressed audio. The host then mentions WMA voice formats at bit rates like 4, 5, and 8 kbps, but focuses on demonstrating what's available in 16 kHz. A warning is given before demonstrating AMR wideband at low bit rates, specifically 4.75 and 12.2 kbps, which produce harsh audio results. The host also includes GSM at 16 kbps and revisits the original waveform for comparison. The episode concludes with additional sampling rate demonstrations at 10 kHz, 12 kHz, and 11,025 Hz. Throughout, the host offers to share original files and compile a more extensive archive, inviting listener feedback for future demos. The format is educational, with a focus on practical audio examples and technical details about compression effects.

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from My Random Podcast