Packet Loss occurs when one or more packets of data travelling across a network fail to reach their destination due to network congestion or failure, faulty hardware or any other reason. In case of real time conversation (Skype/VOIP phone call) over a lossy connection, there isn’t always enough time to resend the packets and the loss is being aggregated. When the stream of arriving packets becomes too slow, a gap is created in the streamed media and the client’s audio player has nothing to play. Each such gap, unless concealed in some way, produces an annoying disturbance. As packet loss increases, it results in unpleasant artifacts such as garbled or muted audio, disturbing lags, pops and glitches. Distortions and audio dropouts are becoming more and more acute and the call might be lost entirely.
PLC (Packet Loss Concealment) technology is designed to mask the effects of the digital voice signal packets loss. The objective of PLC is to reconstruct a synthetic signal in such a way, that it covers the missing data in a received bit stream and thus hides transmission losses. The reconstruction smoothens gaps in the signal and provides a subjective masking of the lost packets, recognizable by the human ear.
Alango PLC algorithm operates with a speech signal in the form of linear PCM audio samples in 16-bit format with typical sampling rate of 8 or 16 kHz. Input-output and signal processing in the PLC algorithm is performed on frames of almost arbitrary length, ranging from a single sample up to dozens of milliseconds. A distinctive feature of Alango PLC algorithm is its ability to recover the audio data from its samples, untied from the boundaries of the received packets.
The initial lossless speech signal, the packet loss signal and the reconstructed (interpolated) signal are shown below:
Use control buttons on the interactive screenshots to play/stop/navigate
Mobile phones, Bluetooth headsets, Hearables, Conferencing equipment.