| Sonstiges: |
- Nachgewiesen in: USPTO Patent Grants
- Sprachen: English
- Patent Number: 7,493,255
- Publication Date: February 17, 2009
- Appl. No: 10/413435
- Application Filed: April 10, 2003
- Assignees: Nokia Corporation (Espoo, FI)
- Claim: 1. Method comprising: calculating linear predictive coefficients from samples of audio signals, for providing a calculated signal indicative of calculated linear predictive coefficients; extracting, in response to said calculated signal, line spectral frequency vectors from said calculated linear predictive coefficients with an extraction rate, for providing a vector signal indicative of extracted linear spectral frequency vectors comprising values of different linear spectral frequency parameters; forming, in response to said vector signal, a track signal indicative of linear spectral frequency tracks for said linear spectral frequency parameters, which linear spectral frequency tracks represent values of respective linear spectral frequency parameters over time; low pass filtering, in response to said track signal, said linear spectral frequency tracks with a predetermined cut-off frequency for providing a low pass filtered signal, wherein less linear spectral frequency tracks than can be formed from said linear spectral frequency parameters are low pass filtered with a predetermined cut-off frequency; and reconstructing, in response to said low pass filtered signal, a decimated number of linear spectral frequency vectors from said low pass filtered linear spectral frequency tracks, said decimated number corresponding to a selected vector output rate, which is lower than said extraction rate.
- Claim: 2. Method according to claim 1 , wherein said linear spectral frequency vectors extracted from said calculated linear predictive coefficients with an extraction rate higher than said selected vector output rate are extracted for all samples of said audio signals from which linear predictive coefficients are calculated.
- Claim: 3. Method according to claim 1 , wherein said linear spectral frequency vectors extracted from said linear predictive coefficients with an extraction rate higher than said selected vector output rate are extracted with an extraction rate which is lower than the sample rate of said audio signals from which linear predictive coefficients are calculated.
- Claim: 4. Method according to claim 1 , wherein an linear spectral frequency track is formed for each of said linear spectral frequency parameters and wherein each of said linear spectral frequency tracks is low pass filtered with a predetermined cut-off frequency.
- Claim: 5. Method according to claim 1 , wherein a dedicated optimal inter-frame predictor is determined for said linear spectral frequency vectors reconstructed with said selected vector output rate from said low pass filtered linear spectral frequency tracks.
- Claim: 6. Method according to claim 1 , wherein an optimized vector quantization codebook is employed for quantizing said linear spectral frequency vectors of said desired vector output rate, which codebook is generated based on a dedicated codebook training for said linear spectral frequency vectors reconstructed with said selected vector output rate from said low pass filtered linear spectral frequency tracks.
- Claim: 7. Method according to claim 1 , wherein said cut-off frequency F is selected dependent on said selected linear spectral frequency vector output rate 1/T according to an equation F≈1/(2*T).
- Claim: 8. Apparatus comprising: a calculation component, responsive to an audio signal, for calculating linear predictive coefficients from samples of said audio signal for providing a coefficient signal indicative of calculated linear predictive coefficients; an extraction component for extracting, in response to said coefficient signal, linear spectral frequency vectors from said linear predictive coefficients with an extraction rate, for providing a vector signal with extracted linear spectral frequency vectors comprising values of different linear spectral frequency parameters; a track forming component for forming, in response to said vector signal, linear spectral frequency tracks for said linear spectral frequency parameters, for providing a track signal with formed linear spectral frequency tracks indicative of values of respective linear spectral frequency parameters over time; a low pass filter component for low pass filtering, in response to said track signal, said formed linear spectral frequency tracks with a predetermined cut-off frequency for providing a low pass filtered signal, wherein less linear spectral frequency tracks than can be formed from said linear spectral frequency parameters are low pass filtered with a predetermined cut-off frequency; and a reconstruction component for reconstructing, in response to said low pass filtered signal, a decimated number of linear spectral frequency vectors from said low pass filtered linear spectral frequency tracks, said decimated number corresponding to a selected vector output rate, which is lower than said extraction rate, for providing a reconstructed signal indicative thereof.
- Claim: 9. Apparatus comprising processing means for: calculating, in response to an audio signal, linear predictive coefficients from samples of audio signals and providing a coefficient signal indicative of calculated linear predictive coefficients; extracting, in response to said coefficient signal, linear spectral frequency vectors from said linear predictive coefficients with an extraction rate, for providing a vector signal with said extracted linear spectral frequency vectors comprising values of different linear spectral frequency parameters; forming, in response to said vector signal, linear spectral frequency tracks for said linear spectral frequency parameters, for providing a track signal with formed linear spectral frequency tracks indicative of values of respective linear spectral frequency parameters over time; low pass filtering said track signal with said formed linear spectral frequency tracks with a predetermined cut-off frequency and providing a low pass filtered signal, wherein less linear spectral frequency tracks than can be formed from said linear spectral frequency parameters are low pass filtered with a predetermined cut-off frequency; and reconstructing, in response to said low pass filtered signal, a decimated number of linear spectral frequency vectors from said low pass filtered linear spectral frequency tracks, said decimated number corresponding to a selected vector output rate, which is lower than said extraction rate, and providing a reconstructed signal indicative thereof.
- Claim: 10. Chip, comprising means for: calculating, in response to an audio signal, linear predictive coefficients from samples of audio signals and providing a coefficient signal indicative of calculated linear predictive coefficients; extracting, in response to said coefficient signal, linear spectral frequency vectors from said linear predictive coefficients with an extraction rate, for providing a vector signal with said extracted linear spectral frequency vectors comprising values of different linear spectral frequency parameters; forming, in response to said vector signal, linear spectral frequency tracks for said linear spectral frequency parameters, for providing a track signal with formed linear spectral frequency tracks indicative of values of respective linear spectral frequency parameters over time; low pass filtering said track signal with said formed linear spectral frequency tracks with a predetermined cut-off frequency and providing a low pass filtered signal, wherein less linear spectral frequency tracks than can be formed from said linear spectral frequency parameters are low pass filtered with a predetermined cut-off frequency; and reconstructing, in response to said low pass filtered signal, a decimated number of linear spectral frequency vectors from said low pass filtered linear spectral frequency tracks, said decimated number corresponding to a selected vector output rate, which is lower than said extraction rate, and providing a reconstructed signal indicative thereof.
- Claim: 11. Communication network comprising an encoder with processing means for: calculating, in response to an audio signal, linear predictive coefficients from samples of audio signals and providing a coefficient signal indicative of calculated linear predictive coefficients; extracting, in response to said coefficient signal, linear spectral frequency, linear spectral frequency vectors from said linear predictive coefficients with an extraction rate, for providing a vector signal with said extracted linear spectral frequency vectors comprising values of different linear spectral frequency parameters; forming, in response to said vector signal, linear spectral frequency tracks for said linear spectral frequency parameters, for providing a track signal with formed linear spectral frequency tracks represent values of respective linear spectral frequency parameters over time; low pass filtering said track signal with said formed linear spectral frequency tracks with a predetermined cut-off frequency and providing a low pass filtered signal, wherein less linear spectral frequency tracks than can be formed from said linear spectral frequency parameters are low pass filtered with a predetermined cut-off frequency; and reconstructing, in response to said low pass filtered signal, a decimated number of linear spectral frequency vectors from said low pass filtered linear spectral frequency tracks, said decimated number corresponding to a selected vector output rate, which is lower than said extraction rate, and providing a reconstructed signal indicative thereof.
- Claim: 12. Communication system comprising a communication network and at least one mobile station, wherein at least one of said communication network and said at least one mobile station comprises processing means for calculating, in response to an audio signal, linear predictive coefficients from samples of audio signals and providing a coefficient signal indicative of calculated linear predictive coefficients; extracting, in response to said coefficient signal, linear spectral frequency vectors from linear predictive coefficients with an extraction rate, for providing a vector signal with said extracted linear spectral frequency vectors comprising values of different linear spectral frequency parameters; forming, in response to said vector signal, linear spectral frequency tracks for said linear spectral frequency parameters, for providing a track signal with formed linear spectral frequency tracks represent values of respective linear spectral frequency parameters over time; low pass filtering said track signal with said formed linear spectral frequency tracks with a predetermined cut-off frequency and providing a low pass filtered signal, wherein less linear spectral frequency tracks than can be formed from said linear spectral frequency parameters are low pass filtered with a predetermined cut-off frequency; and reconstructing, in response to said low pass filtered signal, a decimated number of linear spectral frequency vectors from said low pass filtered linear spectral frequency tracks, said decimated number corresponding to a selected vector output rate, which is lower than said extraction rate, and providing a reconstructed signal indicative thereof.
- Claim: 13. Computer program with program code stored on a computer-readable medium for calculating, in response to an audio signal, linear predictive coefficients from samples of audio signals and providing a coefficient signal indicative of calculated linear predictive coefficients; extracting, in response to said coefficient signal, linear spectral frequency vectors from said linear predictive coefficients with an extraction rate, for providing a vector signal with said extracted linear spectral frequency vectors comprising values of different linear spectral frequency parameters; forming, in response to said vector signal, linear spectral frequency tracks for said linear spectral frequency parameters, for providing a track signal with formed linear spectral frequency tracks indicative of values of respective linear spectral frequency parameters over time; low pass filtering said track signal with said formed linear spectral frequency tracks with a predetermined cut-off frequency and providing a low pass filtered signal, wherein less linear spectral frequency tracks than can be formed from said linear spectral frequency parameters are low pass filtered with a predetermined cut-off frequency; and reconstructing, in response to said low pass filtered signal, a decimated number of linear spectral frequency vectors from said low pass filtered linear spectral frequency tracks, said decimated number corresponding to a selected vector output rate, which is lower than said extraction rate, and providing a reconstructed signal indicative thereof wherein said computer program is for execution by signal processing means.
- Claim: 14. Computer program product with a program code, which program code is encoded on a machine readable storage medium, for calculating, in response to an audio signal, linear predictive coefficients from samples of audio signals and providing a coefficient signal indicative of calculated linear predictive coefficients; extracting, in response to said coefficient signal, linear spectral frequency vectors from said linear predictive coefficients with an extraction rate, for providing a vector signal with said extracted linear spectral frequency vectors comprising values of different linear spectral frequency parameters; forming, in response to said vector signal, linear spectral frequency tracks for said linear spectral frequency parameters, for providing a track signal with formed linear spectral frequency tracks indicative of values of respective linear spectral frequency parameters over time; low pass filtering said track signal with said formed linear spectral frequency tracks with a predetermined cut-off frequency and providing a low pass filtered signal, wherein less linear spectral frequency tracks than can be formed from said linear spectral frequency parameters are low pass filtered with a predetermined cut-off frequency; reconstructing, in response to said low pass filtered signal, a decimated number of linear spectral frequency vectors from said low pass filtered linear spectral frequency tracks, said decimated number corresponding to a selected vector output rate, which is lower than said extraction rate, and providing a reconstructed signal indicative thereof; wherein said program code is for execution by signal processing means.
- Claim: 15. Mobile station for a communication system comprising a calculation component, responsive to an audio signal, for calculating linear predictive coefficients from samples of audio signals and providing a coefficient signal indicative of calculated linear predictive coefficients; a vector extraction component, responsive to said coefficient signal, for extracting linear spectral frequency vectors from said linear predictive coefficients with an extraction rate, for providing a vector signal with said extracted linear spectral frequency vectors comprising values of different linear spectral frequency parameters; a track forming component, responsive to said vector signal, for forming linear spectral frequency tracks for said linear spectral frequency parameters, for providing a track signal with formed linear spectral frequency tracks indicative of values of respective linear spectral frequency parameters over time; a filter component, responsive to said track signal, for low pass filtering formed linear spectral frequency tracks with a predetermined cut-off frequency and providing a low pass filtered signal, wherein less linear spectral frequency tracks than can be formed from said linear spectral frequency parameters are low pass filtered with a predetermined cut-off frequency; and a reconstruction component, responsive to said low pass filtered signal, for reconstructing a decimated number of linear spectral frequency vectors from said low pass filtered linear spectral frequency tracks, said decimated number corresponding to a selected vector output rate, which is lower than said extraction rate, and providing a reconstructed signal indicative thereof.
- Claim: 16. Encoder comprising a calculation component, responsive to an audio signal, for calculating linear predictive coefficients from samples of audio signals and providing a coefficient signal indicative of calculated linear predictive coefficients; a vector extraction component, responsive to said coefficient signal, for extracting linear spectral frequency vectors from said linear predictive coefficients with an extraction rate, for providing a vector signal with said extracted linear spectral frequency vectors comprising values of different linear spectral frequency parameters; a track forming component, responsive to said vector signal, for forming linear spectral frequency tracks for at least one of said linear spectral frequency parameters, which linear spectral frequency tracks represent values of respective linear spectral frequency parameters, for providing a track signal with formed linear spectral frequency tracks indicative of said values of respective linear spectral frequency parameters over time; a filter component, responsive to said track signal, for low pass filtering said formed linear spectral frequency tracks with a predetermined cut-off frequency and providing a low pass filtered signal, wherein less linear spectral frequency tracks than can be formed from said linear spectral frequency parameters are low pass filtered with a predetermined cut-off frequency; and a reconstruction component, responsive to said low pass filtered signal, for reconstructing a decimated number of linear spectral frequency vectors from sad low pass filtered linear spectral frequency tracks, said decimated number corresponding to a selected vector output rate, which is lower than said extraction rate, and providing a reconstructed signal indicative thereof.
- Claim: 17. Chip comprising a calculation component, responsive to an audio signal, for calculating linear predictive coefficients from samples of audio signals and for providing a coefficient signal indicative of calculated linear predictive coefficients; a vector extraction component, responsive to said coefficient signal, for extracting linear spectral frequency vectors from said linear predictive coefficients with an extraction rate, for providing a vector signal with said extracted linear spectral frequency vectors comprising values of different linear spectral frequency parameters; a track forming component, responsive to said vector signal, for forming linear spectral frequency tracks for said linear spectral frequency parameters, which linear spectral frequency tracks represent the value of respective linear spectral frequency parameters, for providing a track signal with formed linear spectral frequency tracks indicative of said values of respective linear spectral frequency parameters over time; a filter component, responsive to said track signal, for low pass filtering said formed linear spectral frequency tracks with a predetermined cut-off frequency and providing a low pass filtered signal, wherein less linear spectral frequency tracks than can be formed from said linear spectral frequency parameters are low pass filtered with a predetermined cut-off frequency; and a reconstruction component, responsive to said low pass filtered signal, for reconstructing a decimated number of linear spectral frequency vectors from said low pass filtered linear spectral frequency tracks, said decimated number corresponding to a selected vector output rate, which is lower than said extraction rate, and providing a reconstructed signal indicative thereof.
- Claim: 18. Communication network comprising an encoder with a calculation component, responsive to an audio signal, for calculating linear predictive coefficients from samples of audio signals and for providing a coefficient signal indicative of calculated linear predictive coefficients; a vector extraction, responsive to said coefficient signal, for extracting linear spectral frequency vectors from said linear predictive coefficients with an extraction rate, for providing a vector signal with said extracted linear spectral frequency vectors comprising values of different linear spectral frequency parameters; a track forming component, responsive to said vector signal, for forming linear spectral frequency tracks for said linear spectral frequency parameters, which linear spectral frequency tracks represent the values of respective linear spectral frequency parameters, for providing a track signal with formed linear spectral frequency tracks indicative of said values of respective linear spectral frequency parameters over time; a filter component, responsive to said track signal, for low pass filtering said formed linear spectral frequency tracks with a predetermined cut-off frequency and providing a low pass filtered signal, wherein less linear spectral frequency tracks than can be formed from said linear spectral frequency parameters are low pass filtered with a predetermined cut-off frequency; and a reconstruction component, responsive to said low pass filtered signal, for reconstructing a decimated number of linear spectral frequency vectors from said low pass filtered linear spectral frequency tracks, said decimated number corresponding to a selected vector output rate, which is lower than said extraction rate, and providing a reconstructed signal indicative thereof.
- Claim: 19. Communication system comprising a communication network and at least one mobile station, wherein at least one of said communication network and said at least one mobile station comprises a calculation component, responsive to an audio signal, for calculating linear predictive coefficients from samples of audio signals and for providing a coefficient signal indicative of calculated linear predictive coefficients; a vector extraction component, responsive to said coefficient signal, for extracting linear spectral frequency vectors from said linear predictive coefficients with an extraction rate, for providing a vector signal with said extracted linear spectral frequency vectors comprising values of different linear spectral frequency parameters; a track forming component, responsive to said vector signal, for forming linear spectral frequency tracks for said linear spectral frequency parameters, which linear spectral frequency tracks represent the values of respective linear spectral frequency parameters, for providing a track signal with formed linear spectral frequency tracks indicative of said values of respective linear spectral frequency parameters over time; a filter component, responsive to said track signal, for low pass filtering said formed linear spectral frequency tracks with a predetermined cut-off frequency and providing a low pass filtered signal, wherein less linear spectral frequency tracks than can be formed from said linear spectral frequency parameters are low pass filtered with a predetermined cut-off frequency; and a reconstruction component, responsive to said low pass filtered signal, for reconstructing a decimated number of linear spectral frequency vectors from said low pass filtered linear spectral frequency tracks, said decimated number corresponding to a selected vector output rate, which is lower than said extraction rate, and providing a reconstructed signal indicative thereof.
- Current U.S. Class: 704/219
- Patent References Cited: 5675701 October 1997 Kleijn et al. ; 5727123 March 1998 McDonough et al. ; 6081776 June 2000 Grabb et al. ; 6275796 August 2001 Kim et al. ; 2002/0055837 May 2002 Ahonen ; 0011649 March 2000
- Other References: “Line Spectrum Representation of Linear Predictor Coefficients of Speech Signals”, F. Itakura, Journal of Acoustic Society of America, vol. 57, p. 535, Apr. 1975. cited by other ; “Efficient Parameter Quantisation for 2.4/1.2 kb/s Split Band LPC Coding”, S. Villette et al, IEEE Workshop on Speech Coding, Dalavan WIS, USA, Sep. 17-30, 2000. cited by other ; “Spectral Dynamics is More Important than Spectral Distortions”, H. P. Knagenhjelm et al, 1995 Conf. on Acoustics, Speech & Signal Processing, Conf. Proceedings, IEEE Part, vol. 1, 1995, pp. 732-735, vol. 1.1 NY USA. cited by other ; “Low-Rate Quantization of Spectral Information in a 4 KB/s Pitch-Synchronous CELP Coder”, D. Guerchi et al, 2000 IEEE Workshop on Speech Coding, Sep. 17-20, 2000, pp. 111-113, Delavan WI. cited by other ; “Digital Signal Processing”, J.G. Proakis et al, (prentice Hall) 1996, paragraph 10.2, pp. 784-787. cited by other ; “Low Rate Quantization of Spectrum Parameters”, T. Eriksson et al, IEEE ICASSP 2000, Jun. 5-9, 2000, Istanbul, Turkey, pp. 1447-1450. cited by other
- Primary Examiner: Chawan, Vijay B
|
