Ling250-15-SpeechTech1-TTS-S16 - 1 Linguistics 250 Sound...

Info icon This preview shows pages 1–3. Sign up to view the full content.

5/26/2016 1 Linguistics 250: Sound Patterns in Human Language May 24,2016 Speech technology: Speech synthesis and text-to-speech systems Today’s topic: Speech synthesis and text-to-speech systems How can we use the principles of sound systems to build flexible, high-quality synthetic speech? What are common approaches, and what are some challenges that arise? 2 Sources: Many of these slides are modified from materials by: Dan Jurafsky (Stanford) Richard Sproat (UIUC) online class materials MITOpenCourseWare: Home > Courses > Electrical Engineering and Computer Science > Automatic Speech Recognition > Lecture Notes Why build synthesis systems? Practical purposes Service or customer-oriented speech generation (e.g. phone systems, transit announcements, Siri et al.) Speech prosthetic devices for people with motor control impairments Theoretical issues Can better understand speech by modeling it What are the critical acoustic components that make sounds intelligible? What are the acoustic features that make it sound natural? How do articulatory configurations translate to acoustic outcomes? 3 What do you need for natural- sounding speech? For any given language, you need: The phoneme structure (the phoneme inventory) The rules of allophony (context dependent rules) The phonotactic rules (permissible syllable structures and possible phoneme combinations) All of these aspects of knowledge of the sounds structure of language: Are intuitive for native speakers of a language Present problems/challenges to non-native speakers Are ALSO an issue for machine generated speech 4 TTS-step 1: Clean text Make the text completely transparent for the computer Expand all abbreviations and symbols- $, Dr., dates, #, etc. Note: this part is easy for us to figure out in context, but not so easy for a computer; e.g. $1999 vs. In 1999 I went to Italy Gilman Dr. vs. Dr. Gilman 12/9/02 in a European vs. American text (December 9 or September 12?) 5 TTS-step 2: Generate a transcription Generate the phonetic transcription of the text (in IPA). Varies in difficulty depending on the orthographic system of the language. English is difficult (irregular spelling)! But specialized dictionaries often contain this information – e.g. the CMU Pronouncing Dictionary 6
Image of page 1

Info icon This preview has intentionally blurred sections. Sign up to view the full version.

5/26/2016 2 TTS step 3: Generate acoustic output Three approaches: Synthesis by rule Control parameters for formant synthesis by specifying fine-grained acoustic details Articulatory synthesis
Image of page 2
Image of page 3
This is the end of the preview. Sign up to access the rest of the document.
  • Spring '14
  • JanetPierrehumbert
  • speech synthesis, Concatenative synthesis, Computer-synthesized sound wave, Articulatory synthesis

{[ snackBarMessage ]}

What students are saying

  • Left Quote Icon

    As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

    Student Picture

    Kiran Temple University Fox School of Business ‘17, Course Hero Intern

  • Left Quote Icon

    I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

    Student Picture

    Dana University of Pennsylvania ‘17, Course Hero Intern

  • Left Quote Icon

    The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

    Student Picture

    Jill Tulane University ‘16, Course Hero Intern