VoiceMatch: A New Offering at NCS

Open PDF
Creating agency: NSA National Security Agency (US)
Document date: 2006-07-18
Type: SIDtoday
Pages: 1
Genre: newsletter
Banner:
Classification: SECRET SI
Dissemination: -

A new four-hour course will teach analysts how use "powerful" new voice-matching technology from MIT/Lincoln Laboratories.

Long Description:
This SIDtoday article announces a new four-hour course (designator TOOL 2019) at the National Cryptologic School teaching analysts to use voice-matching technology obtained from MIT/Lincoln Laboratories. The course covers speech signal basics, the WaveSurfer tool, the Speaker Comparison Algorithm (SCA), model building in the Trainer, and the Speaker Watchlist Tool (SWT). It references earlier voice-matching analyses involving the voices of Bin Laden, Al-Zawahiri, Zarqawi, and other suspected terrorists.
Programs:
VOICEMATCH, SWT, SCA, WAVESURFER

People mentioned:
Cusmariu, Adolf, bin Laden, Osama, al-Zawahiri, Ayman, al-Zarqawi, Abu Musab

Countries:
-

Tags:
speaker identification, training, speech processing, counterterrorism, language analysis

Identifier:
2006-07-18_SIDToday_-_VoiceMatch_A_New_Offering_at_NCS

PDF SHA-256:
5ecf2a2553668ab49240fd4b976ebbd1e51ff8b0e44e24ea19badcb36848dd3f

DYNAMIC PAGE -- HIGHEST POSSIBLE CLASSIFICATION IS
TOP SECRET // SI / TK // REL TO USA AUS CAN GBR NZL

SIDtoday

(C) VoiceMatch: A New Offering at NCS

FROM: Adolf Cusmariu
Technical Director, Operational Technologies (S202B11)
Run Date: 07/18/2006

(S//SI) Several articles this year introduced voice-matching analyses based on powerful
technology recently obtained from MIT/Lincoln Laboratories; the analyses involved the voices of
Bin Laden, Al-Zawahiri, Zarqawi, and other lesser-known or suspected terrorists. (See one of the
earlier articles.)

(C) Due to language analyst interest, the National Cryptologic School has approved a formal
short course (four hours) on practical use of MIT's tools; the designator is TOOL 2019. The
course will be taught by the author, and will be based on a tutorial found at:

tutormit.html

(C) The course will be self-contained and will include some basics of the speech signal to make
the technology more intuitive. Topics will include:

  1. The Speech Signal
  2. The WaveSurfer editing and playback tool
  3. The MIT voice-matching approach
  4. The Speaker Comparison Algorithm (SCA) tool
  5. Model building in the Trainer
  6. The Speaker Watchlist Tool (SWT)
  7. Language-independent features
  8. Error sources
  9. Tricks-of-the-trade

Students will have MIT's tools loaded on their desktop PC's at the NCS, together with a voice
database for comparison and testing. An atmosphere of experimentation and discovery will be
fostered.

(C) In order to give us a better idea how many plan to sign up for this course, please contact
either the author ([REDACTED]) or the NCS's [REDACTED] ([REDACTED]) and let us know of your
interest.

"(U//FOUO) SIDtoday articles may not be republished or reposted outside NSANet
without the consent of S0121 (DL sid_comms)."

DYNAMIC PAGE -- HIGHEST POSSIBLE CLASSIFICATION IS
TOP SECRET // SI / TK // REL TO USA AUS CAN GBR NZL
DERIVED FROM: NSA/CSSM 1-52, DATED 08 JAN 2007 DECLASSIFY ON: 20320108