1 / 19

Nuance Vocalizer for Automotive Nuance Vocalizer Studio

Nuance Vocalizer for Automotive Nuance Vocalizer Studio. SKO-2010 October 2009. Nuance Vocalizer Product Family. Nuance Vocalizer is a family of TTS products built on a scalable engine framework offering solutions with the right features and quality for different target markets

Download Presentation

Nuance Vocalizer for Automotive Nuance Vocalizer Studio

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Nuance Vocalizer for AutomotiveNuance Vocalizer Studio SKO-2010 October 2009

  2. Nuance Vocalizer Product Family Nuance Vocalizer is a family of TTS products built on a scalable engine framework offering solutions with the right features and quality for different target markets Nuance Vocalizer for Mobile Nuance Vocalizer for Automotive Nuance Vocalizer for Network Language and voice development are fully leveraged over all products, all configurable by data files Nuance Vocalizer Studio is a unified toolset supporting the different products of Vocalizer product family

  3. Nuance Vocalizer for Automotive ::A Comprehensive Speech Output Solution NT-Sampa Conversion Tables Vocalizer for Automotive Core Engine with Multiple Voice Models PromptSculptor Prompt Creation Custom Voices Multi-lingual Conversion Tables Recorded Prompt Databases Porting & Integration Audio Postprocessor Standard ~ 20 MB Compact < 2 MB Premium ~ 300 MB User Dictionary Editor Premium Light ~ 60 MB Nuance Vocalizer Studio Example Code and Documentation Services ToolsSuite October2009 Solutions Package Roll-out in 2010

  4. Vocalizer for Automotive : A Total Speech Output Solution • In all applications spoken output is a combination of dynamic and static information. • Vocalizer for Automotive is the single calling point for all speech output and generates speech through seamless combination of : • Pre-recorded audio • Tuned Text-to-Speech • Text-to-Speech • Supported by a unified tuning environment that also allows to customize text processing and optimize phonetics, Vocalizer for Automotive is the most complete and most flexible speech output solution in the market

  5. Vocalizer for Automotive : Key Features • Easy integration of recorded prompts and TTS • No cut-and-paste at application level • Improved prosody of combined prompt + TTS result • Tuned and recorded prompts can be used with same, simple prompt calling mechanism :: <esc>\prompt=nav:001\ • Voices and languages configured by data files only • A single, language independent, core engine simplifies porting and testing efforts, in particular on embedded platforms • Extra voices and languages can be added later in the development process and can be updated more easily, even in deployment • Multiple deployment options / voice models • Low and high footprints and different sampling rates give best quality/footprint options for different platforms • Sharing of optimizations over different deployment options

  6. Vocalizer for Automotive : Add-On’s and Tools • Documentation and sample code for fast ramp-up of application development • Add-on modules • Phonetic mapping tables for NT-SAMPA (NAVTEQ) • Multi-lingual phonetic mapping • Audio post-processing for volume equalization compensating for different audio output hardware • Options for further customization and tuning • User Dictionaries for user domain optimized pronunciations • Regular Expressions for user domain optimized text normalization • ActivePrompts for integration of recorded or tuned prompts

  7. Vocalizer for Automotive : Services • Prompt creation • Prompt recordings are preferred option for highest quality prompts and for large prompt lists • Vocalizer Studio prompt tuning brings maximum flexibility to create smaller prompt lists or to make quick updates to existing deployments without making additional recordings • Prompt tuning can be available as a service but Vocalizer Studio also allows customers to do their own additional tuning • Platform porting and integration support • Porting, platform test and QA • Support of customer integration effort

  8. Highlights Roadmap 2010 • Roll-out of a new high quality voice model for NAVI and infotainment output • Easier and better multi-lingual performance • Extension of functionality in Vocalizer Studio

  9. “Premium Light” : A New Voice Model • “Premium Light” is trained to obtain highest possible quality in NAVI and infotainment domain with special focus on readout of city names, street names and music titles • Using advanced data selection techniques and significantly improved speech compression technology this is achieved in an attractive footprint for high-end automotive projects • The new voice model requires < 64 MB of ROM and < 8 MB of RAM • Available for projects as of OCT-09 • Project driven roll-out of all voices in FY-10

  10. Vocalizer for Automotive : Multi-lingual Support Today • Multi-lingualphoneticinputforstreetandcitynamesissupportedintheVocalizerforAutomotivebythecross-lingualphoneticmapping • Examples • German+Italianextension • BittedemStraßenverlaufinRichtungdas«LagodiGarda»folgen. • French+Germanextension • Dans150mètres,tourneràdroite,danslarue« Dorfstrasse » • Conversiontablesareavailableformostcommonlanguagepairsandcanbecreatedonprojectrequest • ThemethodologycanalsobeappliedtomusictitlesandnamesgeneratedbyCLCVADorCLCMP3module

  11. Vocalizer for Automotive : Multi-lingual Roadmap • Taggingofmulti-lingualinputtextfragmentswillenablemoreflexiblemulti-lingualsupport • phoneticsmappingsandintelligentroutingoflinguisticprocessingwillbedoneinsidetheengine • Advancedmulti-lingualcapabilitieswillbebuiltintonewvoicesfromthestartandaddedtoexistingtopvoices • multilanguagerecordingsintothespeechbase • extendedphoneticset • Examples • German-Italian • BittederAutobahnRichtung«Perugia»folgen • Italian-German • Nelladirezionedi«Heidelberg»,unacodadi6km

  12. Nuance Vocalizer Studio Nuance Vocalizer Studio is an integrated tools environment supporting application development for all products of the Nuance Vocalizer family With Vocalizer Studio you can Evaluate and prototype speechoutput applications Create the highest quality tuned TTS prompts Create user dictionaries with dedicated phonetic transcriptions Create your own recorded prompt databases Design specific text processing rules

  13. Vocalizer Studio : Key Features Vocalizer Studio works directly on the existing Vocalizer for Automotive product installation Immediate availability to tune on all released voices Increased flexibility for voice updates Direct validation of tuning results with the run-time components Full application prototyping with integrated evaluator For ultimate quality Vocalizer Studio offers tuning on a large “development speechbase” and allows “speechbase customization” to bring these tuning results to platform. Separate language data packages are released to enable user dictionary creation

  14. Vocalizer Studio 1.0 : June 2009 Vocalizer Studio 1.0 includes PromptSculptor 4.0 and User Dictionary Editor 1.0

  15. Vocalizer Studio Roadmap Vocalizer Studio 1.1 Oct 2009 Will offer creation of recorded active prompt databases starting from a set of .WAV files Vocalizer Studio 1.2 Jan 2010 Will offer creation and evaluation of RETTT rule sets for text processing and normalization

  16. Vocalizer for Automotive Availability Oct 09 NEW NEW NEW LANGUAGE 35 languages and 48 voices are available for projects !

  17. Vocalizer for Automotive Availability Oct 09 NEW Nuance adds at least 2 new languages and 6 new or refreshed voices every year The 2010 plan anticipates new voices for German, US English, Korean, Italian, … and will choose new languages among Argentinean Spanish, Vietnamese, Malaysian, Bulgarian (*) Only “Compact” version 35 languages and 48 voices are available for projects !

  18. Text Preprocessor Prompt Matcher Linguistic Processor Unit Selector Synthesizer Technical Appendix Slide : Engine Building Blocks and Tuning Options Input Text Speech Language data Voice data Text processing rules User dictionaries ActivePrompt DBs

  19. TechnicalAppendixSlide:VoiceModelSizesandMemoryRequirements Notes: (1) all data for 22 kHz voice models, 16 kHz voice models available on demand (2) further footprint scaling is possible using different audio compression levels (3) size of premium voice model varies largely over different voices (4) depending on memory configuration code may or may not be copied to RAM (5) certain high end hardware platforms might allow further RAM optimizations

More Related