FYI: Summer internships, PhD research
| Author: |
Jim Talley
|
| FYI Body: |
Motorola, Lexicus Division Summer Internships in Speech and Language Processing Lexicus, a Division of Motorola, specializes in handwriting and speech recognition products for the mobile, wireless and desktop markets. The company is a dynamic organization, located in Palo Alto, California, near Stanford University, currently with about 50 employees (but expected to grow well beyond this figure in the near future). For more general background on Lexicus, please visit our web pages at "http://www.mot.com/lexicus/". For our speech (and handwriting) recognition technologies to be successful, it is crucial that our language resources -- pronouncing dictionaries, prepared texts for statistical modeling of (sub-)languages, etc. -- be of unparalleled quality. We are looking to hire a few bright, hardworking, self-starting graduate (or exceptional undergraduate) students as summer interns to help us further refine our English and Mandarin language resources and capabilities. A partial list of possible tasks includes: - Development of methods to transform word pronunciations from a "standard" pronunciation to pronunciations typical of speakers from targeted regions or socio-linguistic groups. - Work towards "representational efficiency" and robustness in dictionary technology -- e.g., production of orthography and pronunciation of derived forms of base words by rule, compounding rules, etc. - Clean-up / verification of existing pronouncing dictionaries. - Development of transducers to map (in both directions) between formatted, punctuated text and the word stream which would be spoken to yield such text. - Development of resources specific to particular domains (e.g., real estate, radiology, inorganic chemistry, etc.). - Tagging of dictionary entries for various properties. - Refinement of capabilities to produce pronunciations for any and all novel words we might encounter (e.g., newly coined words, acronyms, foreign names, etc.) Desirable skills/experience include: - Reasonable programming ability (especially in perl, 'C', or C++ [unfortunately, languages such as Lisp and Prolog are not of much interest]) - Excellent English communication skills. - Solid background in areas of linguistics other than syntax & semantics (i.e., phonology [esp. computational], phonetics, socio-linguistics [dialectology], morphology,...) - Prior exposure to (or strong interest in) ASR, TTS, or other speech technology. - Experience with machine learning, grammar induction, statistics, and/or pattern recognition technology. - Solid background (esp. implementation experience) with formal automata / mathematical linguistics. - Native fluency in Mandarin (for at least one of the available positions). As an intern at Lexicus, you would: - Gain hands-on, industry experience in the applied language technology field (not with toy problems or "concep demonstrations"); - Potentially make a significant contribution to the state of the art in commercial, large vocabulary speech recognition; - Have a chance to work with an exceptional collection of language / recognition technologists in a dynamic organization; and, - Enjoy all of the benefits of living in the Bay Area. If interested, please forward a resume to the following contact by e-mail, fax or snail mail with the title "Lexicus Speech Intern" in the subjec line and/or cover letter: Debbie Mayer, Human Resources Motorola, Lexicus Division 3145 Porter Drive Palo Alto, CA 94304 Tel: (415) 858-6115 Fax: (415) 494-1148 web: www.mot.com/lexicus e-mail: debbiem@lexicus.mot.com - ---------------------------------------------------------------------- Jim Talley, Sr. Research Scientis - ----------------------------------+----------------------------------- Motorola, Lexicus Division | talley@lexicus.mot.com 3145 Porter Drive | (415)858-6112 (voice) Palo Alto, CA 94304 | (415)494-1146 (FAX) - ----------------------------------+----------------------------------- Pager: (800)759-8888 PIN #1959054 (or via e-mail: 1959054@skymail.com) - ---------------------------------------------------------------------- |
