|
Chapter 4 - Speech Recognition/Synthesis Algorithms |
|
Algorithm |
Third Party |
TMS320 Device |
Program Memory |
Data Memory |
Processor Loading (MIPS) |
Language |
Notes |
| Discrete Speaker-Dependent Recognition | |||||||
| VProPRL™ | VPC | 'C3x | 21 | ||||
| Continous Speaker-Independent Recognition | |||||||
| L&H.asr1000/T Automatic Speech Recognition ASR | L&H | 'C2x/'C5x | 64 | 28+16-370/chan | 12 | US English, French, German, Spanish, Dutch, Italian | Telephony |
| L&H.asr1000/T Automatic Speech Recognition ASR | L&H | 'C3x | 1001 | 28+16-370/chan2 | 8 | US English, French, German, Spanish, Dutch, Italian | Telephony |
| VProPRL™ | VPC | 'C3x | 21 | ||||
| Continous Speaker Independent and Language Independent | |||||||
| L&H.asr1500/M Automatic Speech Recognition ASR | L&H | 'C2x/'C3x/'C5x | 7 | Sample 8+11 kHz | |||
| L&H.asr1500/T Automatic Speech Recognition ASR | L&H | 'C2x/'C3x/'C5x | 7 | Telephony | |||
| Text-to-Speech | |||||||
| ELOQUENS® 2000 Speech Synthesis | CSELT | 'C3x | 80 | 820 | 2.4 | Italian | Multi channel |
| L&H.tts2000/T | L&H | 'C31 | 110 | 370-560 | 4 | 7 | Telephony |
| L&H.tts3000/A | L&H | 'C25 | 64 | 220 | 5 | 7 | Sample 10 kHz |
| L&H.tts3000/C | L&H | 'C10 | 8 | 100 | 0.3 | 7 | Sample 10 kHz |
| L&H.tts3000/T | L&H | 'C3x | 110 | 220-400 | 5 | 7 | Telephony |
| Speaker-Independent Recognition | |||||||
| AURIS® 1010 | CSELT | 'C3x | 64 | 256 | 6.4 | Italian | Telephony |
| AURIS® 2010 | CSELT | 'C3x | 100 | 512 | 11.2 | Italian | Telephony |
| FLEXUS® 1000 | CSELT | 'C3x | 100 | 512 | 6.4 | Italian | Telephony |
| L&H.asr200/A Automatic Speech Recognition ASR | L&H | 'C25/'C5x | 7 | 8+5+4.5/word | 10 | English | |
| L&H.asr200/A Automatic Speech Recognition ASR | L&H | 'C3x | 12 | 16+5+4.5/word | 8 | English | |
| VProPRL™ | VPC | 'C3x | 21 |
| 1 Approximate figures excluding drivers, kernels, DSP OS,
etc. 2 Approximate application-dependent figures 
Return to the
Table of Contents |