Sources and licences for the audio on diffio.ai/enhance-voicemail-audio/ Prepared November 20, 2026. Each file is an excerpt of a test voicemail from a 36-clip test: a clean recording sent through real phone codecs on a simulated network (not a real voicemail), the copy Diffio 4.7 Pro cleaned on November 19, 2026, and the clean recording itself. All three are level-matched (one fixed gain each) and saved as mono MP3. No one recorded is a Diffio customer or endorses Diffio. Details: manifest.json. 1. quiet-m4a-phone.mp3, quiet-m4a-diffio.mp3, quiet-m4a-clean.mp3 Voice: Ty Lasky, LibriVox (librivox.org). "Letters on Different Subjects in Natural Philosophy, Volume 1 (Letters to a German Princess), section 68 (More Particular Account of the Dispute Respecting Universal Gravitation)". https://librivox.org/letters-on-natural-philosophy-vol1-by-leonhard-euler Licence of the voice recording: public domain (LibriVox recording; LibriVox dedicates its recordings to the public domain) Changes: sent through phone codecs, cleaned with Diffio, excerpted and level-matched. 2. weak-signal-amr-phone.mp3, weak-signal-amr-diffio.mp3, weak-signal-amr-clean.mp3 Voice: Algy Pug, LibriVox (librivox.org). "Letters to Dead Authors, section 21 (To Omar Kha'yya'm)". https://librivox.org/letters-to-dead-authors-by-andrew-lang/ Licence of the voice recording: public domain (LibriVox recording; LibriVox dedicates its recordings to the public domain) Noise: "DEMAND: a collection of multi-channel recordings of acoustic noise in diverse environments", Joachim Thiemann, Nobutaka Ito, Emmanuel Vincent (2013), https://zenodo.org/records/1227121, CC BY-SA 3.0 (https://creativecommons.org/licenses/by-sa/3.0/). These three files are adaptations of DEMAND and are shared under CC BY-SA 3.0. Changes: noise mixed under the voice, sent through phone codecs, cleaned with Diffio, excerpted and level-matched. 3. car-noise-phone.mp3, car-noise-diffio.mp3, car-noise-clean.mp3 Voice: Jonathan Jones, LibriVox (librivox.org). "Letters from Hell, section 1 (Letter I)". https://librivox.org/letters-from-hell-by-valdemar-adolph-thisted/ Licence of the voice recording: public domain (LibriVox recording; LibriVox dedicates its recordings to the public domain) Noise: "DEMAND: a collection of multi-channel recordings of acoustic noise in diverse environments", Joachim Thiemann, Nobutaka Ito, Emmanuel Vincent (2013), https://zenodo.org/records/1227121, CC BY-SA 3.0 (https://creativecommons.org/licenses/by-sa/3.0/). These three files are adaptations of DEMAND and are shared under CC BY-SA 3.0. Changes: noise mixed under the voice, sent through phone codecs, cleaned with Diffio, excerpted and level-matched. 4. older-man-phone.mp3, older-man-diffio.mp3, older-man-clean.mp3 Voice: Rafael Reyes, Courtesy U.S. National Parks. "Reyes, Rafael_Z35_WAPA-246_WAPA 4170_OralHist_Audio_public.mp3 (War in the Pacific NHP oral history)". https://npgallery.nps.gov/AssetDetail/4a01732e-247e-4a24-8e3f-d42e877d8276 Licence of the voice recording: public domain (NPGallery: "This digital asset is in the public domain") Changes: sent through phone codecs, cleaned with Diffio, excerpted and level-matched. 5. dropouts-phone.mp3, dropouts-diffio.mp3, dropouts-clean.mp3 Voice: elizabethhousworth, LibriVox (librivox.org). "Letters on Different Subjects in Natural Philosophy, Volume 1 (Letters to a German Princess), section 75 (Changes Which May Take Place in the State of Bodies)". https://librivox.org/letters-on-natural-philosophy-vol1-by-leonhard-euler Licence of the voice recording: public domain (LibriVox recording; LibriVox dedicates its recordings to the public domain) Changes: sent through phone codecs, cleaned with Diffio, excerpted and level-matched. The whole test: 30 LibriVox recordings (public domain; LibriVox dedicates its recordings to the public domain, each reader credited in the test record) and 6 U.S. National Park Service War in the Pacific oral histories (public domain, Courtesy U.S. National Parks). Room, car and distance clips (18 of 36) use DEMAND noise, "DEMAND: a collection of multi-channel recordings of acoustic noise in diverse environments", Joachim Thiemann, Nobutaka Ito, Emmanuel Vincent (2013), https://zenodo.org/records/1227121, CC BY-SA 3.0 (https://creativecommons.org/licenses/by-sa/3.0/); those 18 test voicemails and every tool's output made from them are adaptations shared under CC BY-SA 3.0. Handling noise and room responses are our own synthetic code. Real call on the page: "Diving Deeper: Exploring Hawaii's Coral Reefs" (December 2015), NOAA Ocean Service podcast, public domain, from diffio.ai's speech enhancer test. NOAA does not endorse Diffio.