Here is a list of possible sources of Spoken Audio files that might be used for the creation of GPL Acoustic Models.

• Audio Source list:
  • ​Gutenburg audio project
  • ​Wikipedia Spoken Articles
  • CMU:
    • ​CMU Audio Database :
      • ​Microphone array database
      • ​Census (AN4) database
      • ​CMU_SIN Speech in Noise Database
      • ​PDA database
      • ​Resource Management (RM1) database (no wav - Sphinx mfc only)
  • Festvox:
    • ​Festvox databases
      • ​CMU ARCTIC (no restrictions)
      • ​CMU_FAF (Facts and Fables) database
      • ​CMU_SIN database Speech in Noise
      • ​CMU Chaplain (for research only)
    • Diphone Databases
      • ​CMU US KAL diphone
      • ​CSTR UK RAB diphone
    • ldom Databases (Limited Domain)
      • ​time ldom (cmu_time_awb_ldom)
      • ​weather ldom (cmu_weather_awb_ldom)
      • ​Communicator ldom (cmu_com_kal_ldom)
  • ISIP/CAVS Switchboard
    • ​ISIP Abridged Switchboard Audio Database
    • ​full Switchboard Audio Database
  • ​American Rhetoric
  • ​MICASE
  • ​TalkBank ​TalkBank Audio Files (GNU license)
    • ​Switchboard database
  • Hansard Canada (Audio feeds on day of debate)
    • ​House of Commons
    • ​Senate
  • ​MICASE Michigan Corpus of Academic Spoken English
  • ​AUE - alt-usage-english
  • ​Voxeo Telephony Audio Files
  • ​Internet Archive's collection of audio recordings

• Links
  • ​Bookmarks for Corpus-based Linguists
  • ​Corpora and other Language and Speech Data under DICE
  • ​David Lee's Bookmarks for Corpus based Linguistics

Other Possible sources, but with licensing issues:

• ​Buckeye Corpus
• CSLU Speech Synthesis Research Group;
  • ​OGIresLPC 2.1.0 voices (voice data not released yet - only for research/personal use ...)
  • CMU
    • ​Let's Go Speech Dialog Data - (license for research only)
  • Festvox
    • ​CSTR US KED Timit (for research, educational and individual use only)

(Also see Ticket #22)