Method and device for the artificial extension of the bandwidth of speech signals
A method for the artificial extension of the bandwidth of speech signals involves: a) Provision of a wideband input speech signal (s wb i (k)); b) Determination of the signal components (s eb (k)) of the wideband input speech signal (s wb i (k)) required for the bandwidth extension from an extension band from the wideband input speech signal (s wb i (k)); c) Determination of the temporal envelopes of the signal components (s eb (k)) determined for the bandwidth extension; d) Determination of the spectral envelopes of the signal components (s eb (k)) determined for bandwidth extension; e) Encoding of the information for the temporal envelopes and the spectral envelopes, and provision of the encoded information by carrying out the extension of the bandwidth; f) Decoding of the encoded information and generation of the temporal envelopes and the spectral envelopes from the encoded information for the production of a bandwidth-extended output speech signal (s wb o (k)).
1. A method for artificial extension of bandwidth of speech signals, comprising:
encoding by a process comprising:
providing a wideband input speech signal, the wideband input speech signal having an extension band outside of non-extended band;
determining signal components within the extension band of the wideband input speech signal, the signal components being required for bandwidth extension into the extension band of the wideband input speech signal;
determining spectral envelopes of the signal components;
determining temporal envelopes of the signal components, the temporal envelopes being determined independently of the spectral envelopes, without using the spectral envelopes as an input; and
encoding information for the temporal envelopes and the spectral envelopes to produce encoded information for extending the bandwidth; and
decoding the encoded information and reconstructing the temporal envelopes and the spectral envelopes from the encoded information to thereby produce an output speech signal with extended bandwidth, wherein decoding and reconstructing comprise:
producing an excitation signal in a decoder from an input signal transmitted to the decoder;
determining a first correction factor from decoded information of the temporal envelopes and from the excitation signal;
in a first reconstruction, forming reconstructed temporal envelopes by multiplying the first correction factor with the excitation signal;
filtering the reconstructed temporal envelopes to produce pulse responses while filtering; and
in a second reconstruction, forming reconstructed spectral envelopes from the pulse responses and the reconstructed temporal envelopes.
2. The method as claimed in claim 1 , wherein
the signal components are determined by bandpass filtering the wideband input speech signal.
3. The method as claimed in claim 1 , wherein
a quantization of the temporal envelopes and the spectral envelopes is carried out prior to the encoding information for the temporal envelopes and the spectral envelopes.
4. The method as claimed in claim 1 , wherein
determining the spectral envelopes is performed by determining signal powers from spectral subbands of the signal components.
5. The method as claimed in claim 4 , wherein
signal segments of the signal components are produced for determining the signal powers of the spectral subbands, and
a Fast Fourier transform is performed on the signal segments.
6. The method as claimed in claim 1 , wherein
determining the temporal envelopes involves determining signal strengths from temporal signal segments of the signal components.
7. The method as claimed in claim 5 , wherein
determining the temporal envelopes involves determining signal strengths from temporal signal segments of the signal components.
8. The method as claimed in claim 1 , wherein
a modulated narrowband signal with a bandwidth frequency range below a bandwidth frequency range of the extension band of the wideband input speech signal is transmitted to the decoder for the production of excitation signal.
9. The method as claimed in claim 1 , wherein
the excitation signal has harmonics of a fundamental frequency of the input signal transmitted to the decoder.
10. The method as claimed in claim 1 , wherein
the signal components within the extension band of the wideband input speech signal are reconstructed from the reconstructed spectral envelopes.
11. The method as claimed in claim 10 , wherein
a narrowband signal with a bandwidth frequency range below a bandwidth frequency range of the extension band of the wideband input signal is transmitted to a decoder,
the output speech signal is determined by summing the narrowband signal transmitted to the decoder and the reconstructed spectral envelopes, and
the output speech signal is output from the decoder.
12. The method as claimed in claim 1 , wherein
a narrowband signal with a bandwidth frequency range below a bandwidth frequency range of the extension band of the wideband input signal is transmitted to a decoder.
13. The method as claimed in claim 12 , wherein
the bandwidth frequency range of the narrowband signal is within that of the wideband input speech signal, and
the bandwidth frequency range of the narrowband signal is from approximately 50 Hz to approximately 3.4 kHz.
14. The method as claimed in claim 1 , wherein
determining signal components within the extension band, determining temporal envelopes, determining spectral envelopes and encoding information are carried out in an encoder, and
the encoded information is transmitted as a digital signal for decoding purposes.
15. The method as claimed in claim 1 , wherein
the wideband input speech signal has a frequency range between approximately 50 Hz and approximately 7 kHz.
16. The method as claimed in claim 1 , wherein
the extension band of the wideband input speech signal has a frequency range of approximately 3.4 kHz to approximately 7 kHz.
17. The method as claimed in claim 1 , wherein decoding and reconstructing further comprise:
determining a second correction factor from decoded information of the temporal envelopes and from the reconstructed spectral envelopes; and
in a third reconstruction, temporally shaping the reconstructed spectral envelopes by multiplying the second correction factor with the reconstructed spectral envelopes.
18. A device for artificial extension of bandwidth of speech signals comprising:
an encoder device comprising:
a first determination unit to determine signal components within an extension band of a wideband input speech signal;
a second determination unit to determine spectral envelopes for the signal components;
a third determination unit to determine temporal envelopes for the signal components, the temporal envelopes being determined independently of the spectral envelopes, without using the spectral envelopes as an input; and
an encoder to encode the temporal envelopes and the spectral envelopes, and produce encoded information; and
a decoder to decode the encoded information and regenerate the temporal envelopes and the spectral envelopes and produce a bandwidth-extended output speech signal, wherein the decoder comprises:
an excitation signal generator to generate an excitation signal from an input signal transmitted to the decoder;
a first correction unit to determine a first correction factor from decoded information of the temporal envelopes and from the excitation signal, and to form reconstructed temporal envelopes by multiplying the first correction factor with the excitation signal; and
a second correction unit to filter the reconstructed temporal envelopes to produce pulse responses while filtering, and to form reconstructed spectral envelopes from the pulse responses and the reconstructed temporal envelopes.
19. The device as claimed in claim 18 , further comprising:
a third reconstruction unit to determine a second correction factor from decoded information of the temporal envelopes and from the reconstructed spectral envelopes, and to temporally shaping the reconstructed spectral envelopes by multiplying the second correction factor with the reconstructed spectral envelopes.
20. A method for artificial extension of bandwidth of speech signals, comprising:
providing a wideband input speech signal, the wideband input speech signal having an extension band outside of non-extended band;
determining signal components within the extension band of the wideband input speech signal, the signal components being required for bandwidth extension into the extension band of the wideband input speech signal;
determining temporal envelopes of the signal components;
determining spectral envelopes of the signal components independently of the temporal envelopes;
encoding information for the temporal envelopes and the spectral envelopes to produce encoded information for extending the bandwidth; and
decoding the encoded information and reconstructing the temporal envelopes and the spectral envelopes from the encoded information to thereby produce an output speech signal with extended bandwidth, the decoding and reconstructing comprising:
producing an excitation signal in a decoder from an input signal transmitted to the decoder;
determining a first correction factor from decoded information of the temporal envelopes and from the excitation signal;
in a first reconstruction, forming reconstructed temporal envelopes by multiplying the first correction factor with the excitation signal; filtering the reconstructed temporal envelopes to produce pulse responses while filtering; and
in a second reconstruction, forming reconstructed spectral envelopes from the pulse responses and the reconstructed temporal envelopes.
21. The method as claimed in claim 20 , wherein decoding and reconstructing further comprise:
determining a second correction factor from decoded information of the temporal envelopes and from the reconstructed spectral envelopes; and
in a third reconstruction, temporally shaping the reconstructed spectral envelopes by multiplying the second correction factor with the reconstructed spectral envelopes.