A code conversion method for converting first code string data conforming to a first speech coding scheme into second code string data conforming to a second speech coding scheme has the steps of decoding the first code string data to generate a first decoded speech, correcting the signal characteristics of the first decoded speech to generate a second decoded speech, and encoding the second decoded speech in accordance with the second speech coding scheme to generate the second code string data.

Code conversion method and device

In order to improve the accuracy of an excitation source for a band-spreading apparatus and to generate a wide-band signal having no gaps, an a band-widening section generates a prediction coefficient αW of a wide-band speech signal from a prediction coefficient αN of a narrow-band speech signal. An oversampling apparatus oversamples a narrow-band speech signal sndN. An interpolation section generates an adaptive signal excPW of a wide-band speech signal from an adaptive signal excPN of the narrow-band speech signal. A zero-filling section generates a noise signal of a wide-band speech signal from a noise signal excNN of the narrow-band speech signal. A noise addition section adds a noise signal which is a gap of the wide-band speech signal and generates a noise signal excNW. An adder generates an excitation source excW for the wide-band speech signal from the adaptive signal excPW and the noise signal excNW of the wide-band speech signal. A wide-band LPC combining section generates a wide-band speech signal. A band suppression section suppresses a frequency band contained in the narrow-band speech signal within the wide-band speech signal. An adder outputs a wide-band speech signal sndW from the wide-band speech signal and the oversampled narrow-band speech signal.

Apparatus for expanding speech bandwidth

Transcoding of speech coders occurs frequently in systems today as subsystems use different codecs for speech compression. Transcoding is traditionally done by fully decoding speech from the first coder before sending it to the second coder. The CELP speech coders, QCELP 13K and G.723.1, are evaluated under tranditional transcoding situations. An LSP-based transcoder is then designed to partially replace the decoder of the leading coder and the encoder or the following coder. This transcoder is evaluated and compared with the traditional case. Listening tests for speech with flat response show that in the traditional case, distortion is most likely due to both the digital and non-digital parts of the transcoding process. The LSP-based transcoder does not demonstrate a clear-cut improvement in "digital distortion" for flat speech. Listening tests with modified IRS response show that distortion in transcoding performance is instead likely linked with the weighting filter within the coders. Suggestions for improvements are given. Acknowledgments I wish to express my thanks to my supervisors, Sharath Manjunath and Andy DeJaco at Qualcomm Inc., and Gregory Wornell at MIT, for their guidance and help througout this thesis. I would like to thank Qualcomm Incorporated and the MIT VIA Program for sponsoring this thesis and for providing an opportunity to do research in San Diego. I will always cherish this experience. I would like to thank all my friends who have been very supportive and encouraging throughout this whole process. Special thanks to my friends in Impact Campus Fellowship for their prayers. These were timely and effective. I would like to thank my parents and family for all they have done for me all these years. My journey up to MIT and through MIT has been a result of the effort they have expended on my behalf. Thank you very much. Lastly, and most importantly, I would like to thank my Father in heaven, and my Lord and Savior, Jesus Christ for the abundant grace that has been available to me througout this experience. I love You. This thesis is dedicated to you all.

Transcoding between QCELP 13K and G.723.1 CELP speech coders

This paper describes a procedure that reduces the spectral distortion in LPC encoded speech preprocessed by a CVSD coder. In this type of tandem configuration (wide-band/narrowband), the CVSD process introduces extraneous wideband noise and a general broadening of the formant bandwidths. When coupled with the formant distortion introduced by the LPC process, the tandem speech appears buzzy, muffled, and of lower quality than either system considered alone. By low-pass filtering the CVSD speech, on a formant adaptive basis, and narrowing the bandwidths of the primary formants, F1 and F2, the input signal to the LPC synthesizer more closely resembles the original unprocessed signal. This spectral enhancement procedure produces a higher quality CVSD/LPC signal than previously realized.

L. Bergeron

Papers

A spectral enhancement procedure for the wideband/Narrowband tandem