Modified concealing device and method for a speech decoder

• US 6,714,908 B1
• Filed: 12/29/1999
• Issued: 03/30/2004
• Est. Priority Date: 05/27/1998
• Status: Expired due to Term

• Current Assignee: NTT Mobile Communications Network Inc. (Nippon Telegraph and Telephone Corporation)
• Original Assignee: NTT Mobile Communications Network Inc. (Nippon Telegraph and Telephone Corporation)
• Inventors: Naka, Nobuhiko
• Primary Examiner(s): Dorvil, Richemond
• Assistant Examiner(s): Han, Qi
• Application Number:
  US09/462,126
• Time in Patent Office: 1,553 Days
• Field of Search: 704/228, 704/208, 704/214, 704/223, 341/94, 340/825
• US Class Current: 704/228
• CPC Class Codes: G10L 19/005 Correction of errors induced by the transmission channel, if related to the coding algorithm
• CPC Class Codes: G10L 19/12 the excitation function being a code excitation, e.g. in code excited linear prediction [CELP] vocoders
• CPC Class Codes: H03M 7/425 for the decoding process only
• CPC Class Codes: H03M 7/6005 Decoder aspects

Abstract

A speech decoder 10 comprises a decoding processing portion 11 and an amplification process control portion 12. Here, the decoding processing potion 11 is a device for decoding a received coded speech signal (bitstream) BS and outputting a decoded speech signal SP. Additionally, the amplification process control portion 12 monitors the state of occurrence of frame errors in the coded speech signal BS, and when the number of successive frame errors exceeds a predetermined reference frame error number, outputs amplification instructions for a predetermined number of frames after the successive frame errors disappear. As a result, instead of codebook data DCB obtained by a decoding process of the decoding processing portion 11, amplified codebook data DACB are supplied to a synthesis filter portion 17, and is written into the codebook decoder 18 of the decoding processing portion 11 as new original codebook data DCBO. Therefore, after successive frame errors, it is possible to decode normally input coded speech signals BS at a power close to the originally intended power, thus enabling the subjective sound quality of the decoded speech SP to be improved.

Claims

1. A modified concealing device for a code-excited linear prediction (LP) decoder that generates excited signals based on frames of received signals and performs an error concealment process in which excited signals are attenuated when an error is found in a received LP signal frame comprising:

(a) a codebook that has a renewable internal state (t−

1) according to which the codebook outputs, indexed by parameters in a received LP signal frame (t), a code vector which is processed into an excited signal vector (t), the excited signal vector being used to renew the internal state (t−

1) of the codebook to an internal state (t), where t−

1, t, . . . indicate discrete time series;

(b) an amplifier that amplifies the excited signal vector;

(c) a switch that selects either the excited signal vector (t) or the amplified excitation signal vector (t) from the amplifier and supplies it to the codebook for renewal of its internal state (t−

1) to an internal state (t);

(d) a counter that counts a number of successive error frames that contain an error in received LP signals; and

(e) a switch controller that directs the switch, if the number of successive error frames counted by the counter exceeds a predetermined number, to select amplified excitation signal vectors for a time period equal to a predetermined number of frames which time period begins from decoding of a first error-free frame after receiving the successive error frames, whereas otherwise directing the switch to select excitation signal vectors.

2. A modified concealing device according to claim 1, wherein the decoder decodes signals coded by a CELP type speech coder.

3. A modified concealing method for a code-excited linear prediction (LP) decoder. comprising steps of:

(a) receiving frames (t, t+1, . . . ) of linear prediction (LP) signals, where t, t+1, . . . are discrete time series;

(b) generating an excited signal vector (t) according to an internal state (t−

1) in a codebook indexed by parameters in a received LP signal frame (t);

(c) renewing the internal state from the internal state (t−

1) to an internal state (t), using the excited signal vector (t);

(d) checking if a frame (t) contains any errors and, if it does, attenuating the excited signal vector (t);

(e) checking if a number of successive error frames exceed a predetermined number; and

(f) if the number of successive error frames exceeds the predetermined number, in place of the steps (c), renewing the internal state, using amplified excited signal vectors, for a time period equal to a predetermined number of frames which time period begins from decoding of a first error-free frame after receiving the successive error frames.