14

Can anyone point me to opensource databases of speech audio with phonemes labelled, .wav or the like + label files, e.g.

# time, phoneme
0.31200 ao
0.41200 th
0.54200 er
...

I'd prefer English, several speakers, and various noise levels, but let's see what's around.
(There's cmu_*_arctic from CMU festvox; it's pretty old, 2005 — there must be more.)

denis
  • 456
  • 2
  • 11

2 Answers2

4

I am not 100% sure if the following links are useful for you. Please let me know if you are looking for something else.

http://www.iitg.ac.in/ece/emstlab/SRdatabase/introduction.php

http://accent.gmu.edu/howto.php

http://www.signalprocessingsociety.org/technical-committees/list/sl-tc/spl-nl/2012-05/the-rss2015-speech-corpus/

Tasos
  • 4,714
  • 3
  • 20
  • 43
  • Thanks @Anastasios; I don't see phoneme label files so far, will ask the RSS2015 people. (Pros might consider phoneme labelling trivial / impossible / useless, comments ?) – denis Dec 03 '13 at 10:29
  • any update on this? did you reach out denis? – szxk Aug 25 '15 at 05:09
1

There is an auxiliary dataset of phoneme labels for the Librispeech corpus. It can be found here Zenodo: LibriSpeech Alignments.

Jon Nordby
  • 211
  • 1
  • 4