Back to results
US 10,652,120 B2Granted
Voice quality monitoring system
AssigneeDolby Laboratories Licensing Corporation
InventorDoh-Suk Kim
Life of this patent
Estimating the term…
Abstract
This disclosure falls into the field of voice communication systems, more specifically it is related to the field of voice quality estimation in a packet based voice communication system. In particular the disclosure provides methods, computer program products and devices for reducing a prediction error of the voice quality estimation by considering forward error correction of lost voice packets.
Claims18
A method for modifying a statistical metric relating to lost voice packets in a packet based voice communication system, the method comprising:
- receiving a statistical metric, the statistical metric being calculated based on
- a sequence of encoded voice packets received by a decoder and transmitted from one or more end-points in the voice communication system, and on
- one or more lost voice packets which were lost during the transmission from the one or more end-points or discarded due to latency and/or jitter in the transmission,
- receiving recovery data relating to recovered voice packets among the one or more of the lost voice packets, wherein the recovered voice packets are recovered using a forward error correction, FEC, scheme and FEC data relating to the lost voice packets, and wherein the recovery data comprises information pertaining to a recovery factor of each of the recovered voice packets and wherein the recovery factor indicates whether a particular voice packet is fully or partially recovered,
- modifying the statistical metric based, at least in part, on the recovery factor, and
- transmitting the modified statistical metric to a voice quality estimator so as to reduce a prediction error of a voice quality estimate when using the modified statistical metric as input to a voice quality estimating algorithm of the voice quality estimator,
- wherein the recovery factor of a recovered voice packet is calculated using a pre trained mapping function trained for determining a recovery factor based on the used FEC scheme and the FEC data, and
- wherein the pre trained mapping function is pre-trained according to a subjective score given to a pool of speech data.
5 more elements
A method for modifying a statistical metric relating to lost voice packets in a packet based voice communication system, the method comprising:
- receiving data comprising a sequence of encoded voice packets transmitted from one or more end-points in the voice communication system, wherein the encoded voice packets transmitted from the one or more end-points comprise the received sequence of encoded voice packets and one or more lost voice packets which were lost during the transmission from the one or more end-points or discarded due to latency and/or jitter in the transmission and wherein the data further comprises forward error correction, FEC, data relating to the encoded voice packets transmitted from the one or more end-points,
- calculating, based on the received sequence of encoded voice packets, a statistical metric relating to the lost voice packets,
- recovering one or more of the lost voice packets using a FEC scheme and the FEC data relating to the lost voice packets,
- generating a recovery factor for each recovered voice packet, the recovery factor indicating whether a particular voice packet is fully or partially recovered, and
- modifying the statistical metric based, at least in part, on the recovered one or more voice packets, so as to reduce a prediction error of a voice quality estimate when using the modified statistical metric as input to a voice quality estimating algorithm,
- wherein the recovery factor of a recovered voice packet is calculated using a pre trained mapping function trained for determining a recovery factor based on the used FEC scheme and the FEC data, and
- wherein the pre trained mapping function is pre-trained according to a subjective score given to a pool of speech data.
4 more elements
A device for modifying a statistical metric relating to lost voice packets in a packet based voice communication system, the device comprising:
- a receiver configured to:
- receive a statistical metric, the statistical metric being calculated based on a sequence of encoded voice packets received by a decoder and transmitted from one or more end-points in the voice communication system, and on one or more lost voice packets which were lost during the transmission from the one or more end-points or discarded due to latency and/or jitter in the transmission, and
- receive recovery data relating to recovered voice packets among the one or more of the lost voice packets, wherein the recovered voice packets are recovered using a forward error correction, FEC, scheme and FEC data relating to the lost voice packets, and wherein the recovery data comprises information pertaining to a recovery factor of each of the recovered voice packets and wherein the recovery factor indicates whether a particular voice packet is fully or partially recovered, and
- one or more processors configured to:
- modify the statistical metric based, at least in part, on the recovery factor, and
- transmit the modified statistical metric to a voice quality estimator so as to reduce a prediction error of a voice quality estimate when using the modified statistical metric as input to a voice quality estimating algorithm of the voice quality estimator,
- wherein the recovery factor of a recovered voice packet is calculated using a pre trained mapping function trained for determining a recovery factor based on the used FEC scheme and the FEC data, and
- wherein the pre trained mapping function is pre-trained according to a subjective score given to a pool of speech data.
5 more elements
A device for modifying a statistical metric relating to lost voice packets in a packet based voice communication system, the device comprising:
- a receiving stage configured to receive data comprising a sequence of encoded voice packets transmitted from one or more end-points in the voice communication system, wherein the encoded voice packets transmitted from the one or more end-points comprise the received sequence of encoded voice packets and one or more lost voice packets which were lost during the transmission from the one or more end-points or discarded due to latency and/or jitter in the transmission, wherein the data further comprises forward error correction, FEC, data relating to the encoded voice packets transmitted from the one or more end-points,
- a calculating stage configured to calculate, based on the received sequence of encoded voice packets, a statistical metric relating to a number of lost voice packets,
- a FEC recovery stage for recovering one or more of the lost voice packets using a FEC scheme and the FEC data relating to the lost voice packets,
- a recovery factor calculation stage for generating a recovery factor for each recovered voice packet, the recovery factor indicating whether a particular voice packet is fully or partially recovered, and
- a statistical modification stage configured to modify the statistical metric based, at least in part, on the recovered one or more voice packets, so as to reduce a prediction error of a voice quality estimate when using the modified statistical metric as input to a voice quality estimating algorithm,
- wherein the recovery factor of a recovered voice packet is calculated using a pre trained mapping function trained for determining a recovery factor based on the used FEC scheme and the FEC data, and
- wherein the pre trained mapping function is pre-trained according to a subjective score given to a pool of speech data.
4 more elements
Dependencies are inferred from the claim text.