phonon-synthetic-tts

This dataset contains verified generic synthetic TTS audio and manifests. It is not derived from YouTube and does not include personal voice recordings. Owner-clone synthetic voices are excluded.

The Phonon-authored labels and metadata are licensed CC-BY-4.0. Generated audio is included as synthetic training material with provenance fields preserved per row.

No personal, Aqua, VoiceInk, dictation, owner-clone, or frozen eval session audio is included.

Takedown or opt-out contact: elliot@arledge.net.

Generation Provenance

• Included rows: 2963
• Included audio seconds: 10356.434
• Copied audio files: 2502
• Excluded owner-clone rows: 180
• Excluded non-synthetic replay rows: 24
• Source IDs: exact_command_curriculum_v0, exact_form_rescue_v1, python_versions_exact_v0, tcpgen_tts_v1_20260610
• TTS models/engines observed: kokoro-82M, unspecified