VintageVoice — open-source TTS for historical speech patterns: Transatlantic accent, newsreel narrators, Edison-cylinder-era voices (1880s-1960s). Machine learning as cultural heritage preservation. - View it on GitHub