About AMP Lab Projects Downloads Publications People Links
Edward
Lin Personal Homepage:
http://www.andrew.cmu.edu/~eclin |
||
Office: HH 3106 Phone: 412-268-5103 |
Mailing Address: Department of ECE, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213-3890 |
[Research Interests] [Project] [Publications]
Research Focus:
Low power multimedia hardware architectures
Real time, speaker independent speech recognition approaches
Multimedia compression techniques
Areas of Interest:
Low power multimedia architecture designs
Until recently, power efficiency in multimedia devices had largely been ignored for high throughput designs. The sudden growth is largely due to two sources. First, as circuits become faster, heat created the high power dissipation increases the need for expensive cooling systems. The second is in portable devices where battery lifetime is important.
Low Power Variable Length Decoder
With the rise in demand for portable multimedia devices, power has become an important element in variable length decoding. By implementing a custom hardware decoder, an increase in speed and reduction in power is achievable.
For more details on the project please see the Low power VLC Project Page.
Low Power Viterbi algorithm with speech recognition application
Until just recently, Sphinx 3 reached real time decoding on new P4 platforms, but for low power devices such as PDAs processing power is still fairly limited. Creating a custom hardware implementation which is lower in power and decodes faster allows for a wider range of possibilities, such as decoding different languages or machine translation in real time.
For more details on the project please see the Speech-in-Silicon Page.
Conference papers:
This
website is maintained by
Devi Parikh |