Selecting a packet loss concealment procedure

1. A method for selecting a packet loss concealment procedure, the method comprising:

classifying a received audio frame as speech or music by determining a stability of a spectral envelope of signals of received audio frames; and

determining a packet loss concealment procedure based at least partially on the speech/music classification.

2. The method according to claim 1 , wherein determining the stability of the spectral envelope of the signals of the received audio frames comprises comparing the spectral envelope of the signals of at least two received audio frames that are consecutive and were correctly received.

3. The method according to claim 1 , wherein determining the stability of the spectral envelope of the signals of the received audio frames comprises calculating a scalar measurement related to the spectral envelope of the signals of the received audio frames.

4. The method according to claim 1 , wherein classifying the received audio frame as speech or music comprises classifying the received audio frame as music responsive to determining that the spectral envelope of the signals of the received audio frames is stable, and classifying the received audio frame as speech responsive to determining that the spectral envelope of the signals of the received audio frames is unstable.

5. The method according to claim 1 , wherein determining the packet loss concealment procedure comprises selecting a Phase Error Concealment Unit (ECU) method based on sinusoidal phase evolution, responsive to classifying the received audio frame as music.

6. The method according to claim 1 , wherein determining the packet loss concealment procedure comprises selecting a pitch model in response to classifying the received audio frame as speech.

7. The method according to claim 1 , wherein determining the packet loss concealment procedure further comprises evaluating one or more class indicators against corresponding thresholds, the one or more class indicators comprising at least one of: a pitch prediction gain, a tilt of the spectral envelope, a voice activity detector flag, one or more power parameters associated with the signals of the received audio frames, one or more tonality measures of the signals of the received audio frames, a harmonic measure of the signals of the received audio frames, and a spectral complexity of the signals of the received audio frames.

8. An apparatus for selecting a packet loss concealment procedure, the apparatus comprising:

a processor; and

a memory storing instructions that, when executed by the processor, cause the apparatus to:

classify a received audio frame as speech or music by determining a stability of a spectral envelope of signals of received audio frames; and

determine a packet loss concealment procedure based at least partially on the speech/music classification.

9. The apparatus according to claim 8 , wherein the instructions to determine the stability of the spectral envelope of the signals of the received audio frames comprise instructions that, when executed by the processor, cause the apparatus to compare the spectral envelope of the signals of at least two received audio frames that are consecutive and were received correctly.

10. The apparatus according to claim 8 , wherein the instructions to determine the stability of the spectral envelope of signals of received audio frames comprise instructions that, when executed by the processor, cause the apparatus to calculate a scalar measurement related to the spectral envelope of the signals of the received audio frames.

11. The apparatus according to claim 8 , wherein the instructions to classify the received audio frame as speech or music comprise instructions that, when executed by the processor, cause the apparatus to classify the received audio frame as music when the spectral envelope of the signals of the received audio frames is stable and classify the received audio frame as speech when the spectral envelope of the signals of the received audio frames is unstable.

12. The apparatus according to claim 8 , wherein the instructions to determine the packet loss concealment procedure further comprise instructions that, when executed by the processor, cause the apparatus to evaluate one or more class indicators against corresponding thresholds, the one or more class indicators comprising at least one of: a pitch prediction gain, a tilt of the spectral envelope, a voice activity detector flag, one or more power parameters associated with the signals of the received audio frames, one or more tonality measures of the signals of the received audio frames, a harmonic measure of the signals of the received audio frames, and a spectral complexity of the signals of the received audio frames.

13. The apparatus according to claim 9 , wherein the apparatus is a wireless terminal.

14. The apparatus according to claim 8 , wherein the apparatus is a transcoding node arranged to perform transcoding of audio.

15. A non-transitory computer readable-medium storing a computer program comprising program instructions that, when executed by processing circuitry of an apparatus, configures the apparatus to select a packet loss concealment procedure, said program instructions including instructions configuring the apparatus to:

classify a received audio frame as speech or music by determining a stability of a spectral envelope of signals of received audio frames; and

determine a packet loss concealment procedure based at least partially on the speech/music classification.

16. A method performed by an apparatus, the method comprising:

receiving audio frames in a stream of audio frames conveying an encoded audio signal;

dynamically classifying the encoded audio signal as being speech or music, based on evaluating a stability of a spectral envelope of the encoded audio signal; and

employing, for a lost or corrupted audio frame in the stream of audio frames, a packet loss concealment procedure that is either tailored for speech or tailored for music, in dependence on whether the encoded audio signal currently is classified as being speech or music, according to the dynamic classification.

17. The method of claim 16 , wherein dynamically classifying the encoded audio signal as being speech or music includes applying hysteresis to limit how quickly a reclassification from music to speech, or vice versa, can occur.

18. The method of claim 16 , wherein evaluating the stability of the spectral envelope of the encoded audio signal comprises evaluating the spectral envelope of the encoded audio signal over two or more received audio frames that are consecutive and were received correctly.