Speech Recognition Tools

Mini-Course Taught May-June, 2005

Lectures

• Course Information (tex)
• Lecture 1: Speech Databases, Transcriptions, and Transcription Tools (tex)
• Lecture 1: Phonetic and Prosodic Transcription Standards (tex)
• Lecture 2: Regular Expressions, Language Modeling, and Perl (tex)
• Lecture 3: Acoustic Features; Speech Recognition Software on the Web (tex)
• Lecture 4: Support Vector Machines and Phonological Distinctive Features (tex)
• Lecture 5: Training Hidden Markov Models using HTK (tex)
  • Scripts: radio8k_scripts.tgz
• Lecture 6: HMM Refinement (tex)
  • Data: 12hour_data.tgz
  • Scripts: 12hour_scripts.tgz
• Lecture 7: Finite State Transducers, Language Models, and Speech Recognition Search(tex)
  • Speech Understanding Example: query.tgz
  • N-Gram Example: ngram.tgz
  • Dictionary Example: dict.tgz
• Eric Fosler-Lussier's FSM Tutorial
  • Data: crypto.tar.gz
• Lecture 8: Using HTK Models with the ATT drecog program; Recognition and Training as FSM Composition (tex)
  • Examples: lecture8_scripts.tgz

Links

• Acoustic and Phonetic Feature Processing
  • PVTK: Three command-line functions that transform, extract, and apply SVMs to periodic vectors.
  • SpeechLib: A library of types and functions that might be useful for speech processing applications.
  • speechfileformats: Matlab toolkit for reading and writing speech file formats.
• Transcription
  • Transcription Tools: perl tools for manipulating various transcription formats
  • File Formats scripts for matlab
  • Praat
  • sox
  • NIST tools, including SPHERE
  • VoiceBox
• Data
  • 8kHz Radio Speech Corpus
  • 8kHz Radio Speech Corpus
  • TOBI-labeled Switchboard files (Yoon, Chavarria, Cole and Hasegawa-Johnson, ICSLP 2004), plus HTK configuration, MLF, and scripting files.
  • Switchboard Transcriptions: normally available here, but since that site is currently down, we are making the transcriptions available here.