⭐ 欢迎来到虫虫下载站! | 📦 资源下载 📁 资源专辑 ℹ️ 关于我们
⭐ 虫虫下载站

📄 readme.txt

📁 lpc 2400 bps语音编解码程序
💻 TXT
字号:
	LPCSIM - Low rate voice coder
	Version 54  13 March 1991

   The lpcsim voice coder includes an LPC analyzer and synthesizer which
convert between LPC parameters (pitch, voicing, energy, and spectrum)
and speech, and coding routines that quantize the parameters at rates
of 2400 bits per second and below.  It reads and writes speech data,
LPC parameters, or bitstream data, using named files or stdin/out.  Its
I/O and coding options are controlled by command line switches.

 Usage: lpcsim ifile ofile
   [-is/ip/ib]   - input speech/params/bits
   [-os/op/ob]   - output speech/params/bits
   [-pb]         - binary parameter file
   [-l #]        - verbosity level (# = 0-6)
   [-order #]    - LPC order
   [-q 0/2400]   - quantization rate

OPTIONS:

* ifile, ofile are the input and output filenames. If ifile is '-' input
  is taken from stdin, and if ofile is '-' output is sent to stdout.
  If ofile is not present, no output is produced.

* -is, -ip, -ib:  input is speech, LPC parameters, or bitstream.
  -os, -op, -ob:  output is speech, parameters, or bitstream.

    Speech data consists of headerless files of 16 bit two's-complement
    linear samples at 8 kHz sample rate.

    LPC parameters are in text format, one line per frame, in 14 columns:
    Energy in dB, 2 half-frame voicing decisions (0=unvoiced, 1=voiced),
    pitch lag in samples, and 10 reflection coefficients.

    Bitstream data is in hexadecimal format, 1 line per frame; not
    applicable for unquantized parameters.  Bit 1 is the rightmost
    (least significant) bit and bit 54 is the leftmost.

    The input and output options determine whether the analyzer, synthesizer,
    both, or neither are run.  This is useful for speeding up processing,
    or limiting debugging data to just the analyzer or synthesizer.

* -pb: LPC parameters are in binary rather than text format. This is an
    efficiency hack for realtime processing.  The -pb switch has no effect
    on speech or bitstream format.

* -l #: This switch (lower-case L) controls the amount of debugging data
    printed, in increasing levels of verbosity from 0 (none) to 6 (dump
    all internal buffers).  The default value of 1 prints processing
    options and statistics.

* -order #: controls the LPC filter order. This has not been tested at
    other than its default value of 10, and will definitely not work if
    parameters are quantized.

* -q: Parameter quantization option.  Recognized values are 0 (unquantized),
    and 2400 (LPC-10e 2400 bps scalar quantization at 54 bits per frame).


EXAMPLES

 lpcsim foo.spd foo_2400.spd
    Use defaults for all processing options: speech in and out, 2400 bps
    quantization, print processing summary only.

 errgen -h 54 -e 1.0 foo.lpb | lpcsim -ib - - | play -n
    Add 1% random errors to 2400 bps bitstream file 'foo.lpb', pipe the
    corrupted bitstream file into lpcsim, and listen to the results.
    (Play -n refers to a locally modified version of the Sun Sparcstation
    play command that can play 16 bit samples.  The lpc simulator runs
    about 50% of realtime on a Sparcstation-2 for speech in and out, so
    realtime listening is possible on a lightly loaded machine.)

⌨️ 快捷键说明

复制代码 Ctrl + C
搜索代码 Ctrl + F
全屏模式 F11
切换主题 Ctrl + Shift + D
显示快捷键 ?
增大字号 Ctrl + =
减小字号 Ctrl + -