Recovery of missing speech packets using the short-time energy and zero-crossing measurements

Application of short time energy analysis in monitoring the stability of arc sound signal

“…Recently, short time energy analysis was employed to analyze the high-pitched speech signals, and presented better performance than conventional linear prediction analysis [9]. Also, it can be used to recovery of missing speech packet [10], and discriminate voiced and unvoiced segments of speech signals [11]. Hence, it can be considered as an effective tool to analyze the acoustic signal in practice.…”

“…In general, the distribution of the short time energy can be clearly presented by histogram of possibility of short time energy. According to relative analysis [10], more stable signal corresponds higher possibility of small values of the short time energy appearing; if the signal is unstable, large value of the short time energy is easy to appear, which means the distribution of the possibilities is more unfocused. Though the possibility can show the distribution information of the short time energy, it cannot intuitively provide the overall distribution situation of the signal under a certain range.…”

Enhanced VoIP by Signal Reconstruction and Voice Quality Assessment

“…In (Aoki, 2004) the proposed reconstruction technique takes account of pitch variation between the previous and the next known signal frames. In (Erdol et al, 1993) two reconstruction techniques are proposed based on slow-varying parameters of a voice signal: short-time energy and zero-crossing rate (or zerocrossing locations). The aim is to ensure amplitude and frequency continuity between the concealment waveform and the lost one.…”

<title>Speech-property-based FEC for Internet telephony applications</title>

“…The first experiment 7 we carry out is to measure the resynchronization time of the decoder after a number of consecutive frames are lost. We vary the position of the frame loss to cause a number of consecutive voiced/unvoiced frames to be lost and then count the number of the following frames until the signal-to-noise ratio (SNR) exceeds a certain threshold.…”

“…The redundant information can be generated either by generic Forward Error Correction (FEC) codes or by exploiting speech-specific properties. The amount of speechspecific information can range from a simple pitch period measurement in the AP/C scheme ( [20]) 1 over short-term energy and zero crossings to recover the basic envelope of the speech signal ( [7]) up to running entire speech coders in parallel with an offset in time ( [8]). Typically such schemes are complemented by concealment schemes, which try to recover missing speech segments from already decoded PCM samples ( [19] …”