90 likes | 232 Views
Evaluation of Speak. Project 2b Due: March 21 st , in class. Overview. Experiments to evaluate performance of your VoIP application (Project 2) Focus not only on how your software performs, but also design, implementation of experiments analysis writeup. Details. Two parts
E N D
Evaluation of Speak Project 2b Due: March 21st, in class
Overview • Experiments to evaluate performance of your VoIP application (Project 2) • Focus not only on how your software performs, but also • design, implementation of experiments • analysis • writeup
Details • Two parts • In-depth user study with friend (called “group”) • More forgiving users, less objective • Briefer user study outside group • For both parts, evaluate by user perception • Brief, 0.5-2 minute conversation • Record user opinions (1-10, or similar) • Pencil and paper, software, …
In Group • Baseline case: 0% loss, UDP, no speech detection, 40ms • Compare versus: • Loss: 0%, 1%, 5%, 10%, 20% • Latency (by increasing the sample interval): 40ms, 100ms, 250ms, 500ms, 1000ms • Connection type: TCP, UDP • Speech detection: on, off • Pick (and justify) perceptual quality measure
Outside Group • 1-3 people outside of your group • Record some basic information • age, gender, profession, VoIP use… • Loss rates of 0%, 5%, 20% • Latencies of 40ms, 500ms, 1000ms • Record • Quality score (scale 1-10, or similar) • Subjective comments ("sounded good", "was hard to talk", etc.). • Ask to compare with land-line phone call
Report • Introduction (brief) • Motivation and Hypotheses • Background on your software (as needed) • Design of your experiments (brief, since given) • Details on above • Analysis (detailed) • Conclusions (brief) • Summarize findings • Abstract • 1 paragraph • Write last, goes first
Hints • See Hints for Project 1b • Graphing • Refer to grading comments from Project 1b • General comments will hold for this report, too
Hand In • Hardcopy! • Due at beginning of class
Grading • 8% Abstract • 7% Introduction • Background (as needed) • 20% Experiment design • In group (1/2) • Out of group (1/2) • 60% Results and analysis • In group (1/2) • Out of group (1/2) • 5% Conclusions • Other • Writing, clarity, errors, etc. (can lose points) (Grading Rubric on Web Page)