Posts

• Converting Ó Raghallaigh (2010)

  Re-implementation of the 'global' phonetiser from Brian Ó Raghallaigh's Ph.D. thesis

  May 16, 2021

• Clarin-Studio Polish Train mono 1-30

  First 30 steps of training

  May 16, 2021

• Clarin-Studio Polish GMM

  Training on Kaggle

  May 15, 2021

• Extract pre-built Kaldi on Kaggle

  For use with wav2vec-u

  May 15, 2021

• G2P with MFA

  Grapheme-to-phoneme conversion with Montreal Forced Aligner (on Kaggle)

  May 14, 2021

• Title

  May 13, 2021

• Azure speech recognition for Irish

  Authentication is a bit of a pain; recognition is fantastic, but let down by number inverse normalisation

  May 4, 2021

• Azure speech recognition for Irish, part 2

  json output contains nbest, timestamps, and text without replacements

  May 4, 2021

• Azure ASR's JSONL output to JSON

  Just getting 1-best

  May 4, 2021

• Playing with auditok

  Original on kaggle

  May 3, 2021

• rclone and Sharepoint

  tl;dr - rclone can't handle 'public' sharepoint folders

  May 2, 2021

• Two speechbrain speech enhancement models

  Quick comparison of two models

  Apr 30, 2021

• Doegen recordings scraper

  Just the scraper, processing later

  Apr 29, 2021

• Praat via parselmouth

  Mostly playing with the examples

  Apr 24, 2021

• BuNaMo to json

  Possibly incomplete; Kaggle version here

  Apr 24, 2021