AMR - Ericsson Network;AMR - Ericsson NetworkAMR - Ericsson Network
Media Manager Mail Access Barbara Hohlt and Steve Czerwinski UC Berkeley Ericsson Presentation 2000.
-
Upload
gervais-shelton -
Category
Documents
-
view
216 -
download
0
Transcript of Media Manager Mail Access Barbara Hohlt and Steve Czerwinski UC Berkeley Ericsson Presentation 2000.
Media Manager Mail Access
Barbara Hohlt and Steve Czerwinski
UC BerkeleyEricsson Presentation 2000
Project Overview
• Make messages more accessible– Enable faster browsing of many voicemails– Access from different devices with different
capabilities
• Media Mail services– Universal Inbox infrastructure– Voicemail is email encoded in MIME
• Transcoding services– Enhance voicemail interaction– Includes: skimmed audio, transcript,
text/audio summary, and outline
Related Work
• SpeechSkimmer, MIT’s MultiMedia Lab [Aaron95]
• Auto-Summarization, Microsoft Research
• CueVideo, IBM• Other Universal Inboxes• IBM’s universal inbox type project
Architecture
Transcoder Service•Voicemail->Text Transcript
•Voicemail->Text Summary
•Voicemail->Text Outline
•Email ->Plain Audio
•Voicemail->Audio Summary
•Voicemail->Skimmed Audio
Mail Access Interface
NinjaMail
Mail Access Interface
POP
Mail Access Interface
IMAP
Media Manager Interface
Media Manager Service
Client
Client
Client
Folder Store
Bhaskar’s Cell-Phone
Barbara’s Desktop
Preference Registry
Automatic Path Creation Service
800-MEDIA-MGR UID: [email protected]
Naming Service
11
mediamgr: Cluster locn.
22
MediaManager Mail Access Service
33
Bhaskar’s PSTN Phone Universal Inbox
Clients
Messages and Content Objects
MessageMedia Reference IDQueue of Content Objects
Content Object Content IDData
Media Reference IDMessage PartContent Type
MediaManagerServiceIF
• getFolders()• getFoldersAs()• getList()• getListAs()• getMessage()• getMessageContent()
Audio Tools• Speech Recognition/Synthesis
– Transcribe voicemail to text– IBM ViaVoice SDK and custom audio libs
• Natural Language Processing– Directed word spotting by “understanding”
content– ViaVoice SRCL
• Pitch – Detecting important words by emphasized pitch
• Pause– Compression through pause removal
• Spurts– Retrieve sentence structure of voicemail
Results
• Pause detection– Worked well for given applications– Playback speedup by 50-70%
• Pitch detection– Problems due to high pitch sounds and
transitions
• Speech recognition– Performance decrease in conversational
settings
• Natural Language Processing– Performed well with small grammar
Example: Adding GSM
• Adding GSMAudio and GSMSummary• Add ContentObject definition to
MediaManager– Add method createObject()– Add cases to Media Manager
• Add method toGSM() to Transcoder• Edit .config file
– External.transcoder.gsm rungsm
• Edit related transcoders– speechSynthesizer and audioSummary()
Implementing Other Mail Stores• Implement MailAccessIF
– MMgetFolders()– MMgetFoldersAs()– MMgetList()– MMgetListAs()– MMgetMessage()– MMgetMessageContent()
• Add new protocol to FolderStore
Conclusion