A noninvasive

single-ended objective quality assessment method called "FSPAV"

which only utilized IP flow features to predict the perceived quality of VoIP service （PQoS） was proposed. The cores of the method are three PQoS related metrics of flow features. It only needs to monitor IP packets containing voice data from the peer user received by a local end user to calculate metrics

with no need to synchronize clocks or parse application protocols. A measure of the individual truth grad is used to map the measurement values of three metrics on a call segment into a single speech qual-ity prediction. During the calculation

the goodness grad for each flow feature can also be derived. Talk experiments on the Internet by using VoIP software QQ and Skype was presented. The experimental results show that the objective as-sessments correlate well with subjective ones

which indicates that the method proposed here is effective.