![background image](http://html.mh-extra.com/html/texas-instruments/msp53c691/msp53c691_user-manual_1097147105.webp)
Recording
E-5
Guidelines for Optimal TI Speech
-
Microphone Preamplifier—Choose a good microphone preamplifier (such
as Martech).
-
Other Audio Equipment—Consider a good analog-to-digital converter
(such as Lucid), a PCI card, a mixer for analog monitoring and routing
(such as Mackie), and a good playback system (such as Mackie).
Recording Session—Important considerations for the recording session
include microphone placement, audio input levels, voice talent direction, and
data capture techniques.
-
Microphone Placement—A free-standing microphone must be approxi-
mately eight inches from the voice talent’s lips. The microphone must be
pointed toward the voice talent’s chin. Use a shockmount to prevent micro-
phone vibration. The voice talent must be instructed not to move his/her
head during the recording since varying the distance from the microphone
results in varying amplitude (energy level) in the signal.
-
Input levels—Set audio input levels to prevent clipping the signal during
recording (anywhere from around –3 dB to –12 dB, depending on the con-
tent to be recorded). Do not apply equalization, filtering, etc. at this point,
since the aim is to get a very
clean unclipped signal.
-
Voice Talent Direction—The voice talent must be instructed to avoid extra-
neous noise (such as page shuffling, scratching, etc.) while speaking. Ask
the voice talent to practice reading the script while levels are being set and
other final preparations are being made. Be sure that final consonants
(such as [m,n] and [b, d, g]) are articulated distinctly. Answer questions
and adjust speaking rate and vocal effort if necessary, correct mistakes
and give some initial direction regarding the number of
takes and how
each take must be said (for example, Take 1–Energetic, Take 2–Clear ar-
ticulation, etc.). Maintain uniform pitch and loudness levels within and
across recording sessions.
Data Capture—Record using a sampling rate of at least 44.1 kHz or 48 kHz.
Save the recorded speech data as 16-bit, mono, signed integer files of the
following type:
.WAV
Windows wave format (preferred)
During the Recording Session—Sample different takes to preprocess and
analyze with the SDS6000.
Summary of Contents for MSP53C691
Page 1: ...MSP53C691 Speech Synthesizer December 2000 MSDS Speech User s Guide SPSU020 ...
Page 22: ...2 10 ...
Page 96: ...C 6 ...