


PS_ANA(1-ESPS)                              ENTROPIC RESEARCH LAB



NNNNAAAAMMMMEEEE
     ps_ana - pitch-synchronous linear-prediction analysis

SSSSYYYYNNNNOOOOPPPPSSSSIIIISSSS
     ppppssss____aaaannnnaaaa [ ----PPPP _p_a_r_a_m__f_i_l_e ] [ ----ffff _o_u_t__f_0_f_i_l_e ] [ ----{{{{pppprrrr}}}} _r_a_n_g_e ] [
      ----ssss _r_a_n_g_e ] [ ----oooo _o_r_d_e_r ] [ ----SSSS _f_r_a_m_e__s_t_e_p ] [ ----iiii _f_r_a_m_e__s_t_e_p ]
     [ ----cccc _s_p_e_c__r_e_p ] [ ----mmmm _s_p_e_c__m_e_t_h_o_d ] [ ----wwww _w_i_n_d_o_w ] [ ----eeee _p_r_e_e_m_-
     _p_h_a_s_i_s ] [ ----zzzz ] [ ----xxxx _d_e_b_u_g__l_e_v_e_l ] iiiinnnn____ddddaaaattttaaaa iiiinnnn____ppppuuuullllsssseeee oooouuuutttt____ffffiiiilllleeee

DDDDEEEESSSSCCCCRRRRIIIIPPPPTTTTIIIIOOOONNNN
     _P_s__a_n_a performs a pitch-synchronous LP  (linear  prediction)
     analysis.   The input file _i_n__d_a_t_a is a speech signal stored
     in a FEA_SD file.  The input file _i_n__p_u_l_s_e,  typically  gen-
     erated  by  the  _e_p_o_c_h_s(_1-_E_S_P_S) program, is a FEA_SD(5-ESPS)
     file that contains pitch pulse locations of _i_n__d_a_t_a.  Either
     _i_n__d_a_t_a  or _i_n__p_u_l_s_e, but not both, may be "-", for standard
     input.

     The output file _o_u_t__f_i_l_e is a FEA file that  contains  spec-
     tral  parameters  of  the  specified type.  Analysis results
     such as signal RMS, F0, and voicing state can optionally  be
     stored  in  the  _o_u_t__f_0_f_i_l_e indicated by the -_f option.  The
     pitch-synchronously  derived  quantities  in   _o_u_t_f_i_l_e   and
     _o_u_t__f_0_f_i_l_e  are periodically resampled at 200 Hz, unless the
     frame step size is otherwise specified using the  ----SSSS  or  ----iiii
     option.  In any case the periodic resampling is performed by
     choosing the values of the pitch-synchronous parameters from
     the  glottal epoch closest to the periodic sampling instant,
     rather than by interpolation.

OOOOPPPPTTTTIIIIOOOONNNNSSSS
     The following options are supported:

     ----PPPP _p_a_r_a_m__f_i_l_e [params]
          uses the parameter  file  _p_a_r_a_m__f_i_l_e  rather  than  the
          default, which is _p_a_r_a_m_s.

     ----ffff _o_u_t__f_0_f_i_l_e
          The output file  name  of  a  file  to  store  analysis
          results  of signal RMS, F0 (fundamental frequency), and
          voicing state.  These results are  periodically  resam-
          pled at the same rate as those in the "out_file".

     ----rrrr _f_i_r_s_t:_l_a_s_t

     ----rrrr _f_i_r_s_t:+_i_n_c_r
          Determines the range of points from input file.  In the
          first form, a pair of unsigned integers gives the first
          and last points of the range.  If _f_i_r_s_t is  omitted,  1
          is  used.   If  _l_a_s_t  is omitted, the last point in the
          file is used.  The second form  is  equivalent  to  the
          first  with  _l_a_s_t  =  _f_i_r_s_t  +  _i_n_c_r.   If the range is



        Copyright Entropic Research Laboratory, Inc. 1G1        1






PS_ANA(1-ESPS)                              ENTROPIC RESEARCH LAB



          omitted, the entire file is processed.

     ----pppp    Same as the ----rrrr option.

     ----ssss _f_i_r_s_t:_l_a_s_t

     ----ssss _f_i_r_s_t:+_i_n_c_r
          Same function as the ----rrrr option, but specifies the range
          of input data in seconds.

     ----oooo _o_r_d_e_r [0]
          Order for LP analysis.  The default value 0  means  the
          program  will  choose  an order equal to _2 + (_F_s/_1_0_0_0),
          where _F_s is the sampling frequency of  the  input  data
          file _i_n__d_a_t_a.

     ----SSSS _f_r_a_m_e__s_t_e_p [_0._0_0_5 * _s_a_m_p_l_i_n_g _f_r_e_q_u_e_n_c_y]
          Specifies frame step in samples.

     ----iiii _f_r_a_m_e__s_t_e_p [_0._0_0_5]
          Specifies frame step in seconds.

     ----cccc _s_p_e_c__r_e_p [RC]
          Spectral  analysis  type.   Valid  value  is  "_R_C"  for
          reflection  coefficients,  or  "_A_F_C" for autoregressive
          filter  coefficients  (or  linear  prediction   coeffi-
          cients).

     ----mmmm _s_p_e_c__m_e_t_h_o_d [AUTOC]
          Method to compute _s_p_e_c__r_e_p.  Valid value is "_A_U_T_O_C" for
          autocorrelation method, or "_C_O_V" for covariance method.

     ----wwww _w_i_n_d_o_w [HANNING]
          Windowing  technique  used  in  the  spectral  analysis
          method.   Valid  value is "_H_A_N_N_I_N_G", "_H_A_M_M_I_N_G", "_R_E_C_T",
          or "_C_O_S_4", for Hanning, Hamming, rectangular, and  cos4
          window, respectively.

     ----pppp _p_r_e_e_m_p_h_a_s_i_s [0.95]
          First-order preemphasis filter constant used before the
          LPC analysis.  Note that the RMS values returned in the
          optional _o_u_t__f_0_f_i_l_e are based on the preemphasized sig-
          nal in the pitch-synchronous analysis window.

     ----zzzz    This option reverses the signs of output  LPC  parame-
          ters in _o_u_t_f_i_l_e.

     ----xxxx _d_e_b_u_g__l_e_v_e_l [0]
          If _d_e_b_u_g__l_e_v_e_l is  positive,  _p_s__a_n_a  prints  debugging
          messages  and  other  information on the standard error
          output.  The messages proliferate  as  the  _d_e_b_u_g__l_e_v_e_l
          increases.   If  _d_e_b_u_g__l_e_v_e_l  is  0  (the  default), no



        Copyright Entropic Research Laboratory, Inc. 2G2        2






PS_ANA(1-ESPS)                              ENTROPIC RESEARCH LAB



          messages are printed.

EEEESSSSPPPPSSSS PPPPAAAARRRRAAAAMMMMEEEETTTTEEEERRRRSSSS
     The following ESPS parameters have the same meanings as  the
     command-line options.

     _s_t_a_r_t - _i_n_t_e_g_e_r

          The first point in the input sampled data file that  is
          processed.   A  value  of 1 denotes the first sample in
          the file.  This is only read if the ----pppp  option  is  not
          used.   If it is not in the parameter file, the default
          value of 1 is used.

     _n_a_n - _i_n_t_e_g_e_r

          The total number of data points to process.  If _n_a_n  is
          0, the whole file is processed. _N_a_n is read only if the
          ----pppp option is not used.  (See  the  discussion  under  ----
          ppppffffPPPP))))....

      _f_r_a_m_e__s_t_e_p - _f_l_o_a_t
          Analysis frame step interval to use when resampling the
          output parameters.  Valid values are in the range [1/Fs
          0.1].

     _o_r_d_e_r - _i_n_t

          See OPTIONS.

     _s_p_e_c__r_e_p - _s_t_r_i_n_g

          See OPTIONS.

     _s_p_e_c__m_e_t_h_o_d - _s_t_r_i_n_g

          See OPTIONS.

     _w_i_n_d_o_w - _s_t_r_i_n_g

          See OPTIONS.

     _p_r_e_e_m_p_h_a_s_i_s - _f_l_o_a_t

          See OPTIONS.

     _p_e_r__s_t_e_p - _f_l_o_a_t

          Step size in seconds for periodic analysis in  unvoiced
          regions.

     _p_e_r__w_s_i_z_e - _f_l_o_a_t



        Copyright Entropic Research Laboratory, Inc. 3G3        3






PS_ANA(1-ESPS)                              ENTROPIC RESEARCH LAB



          Window duration in seconds  for  periodic  analysis  in
          unvoiced regions.

     _p_s__w_s_i_z_e - _f_l_o_a_t

          Window duration in fraction  of  voiced  pitch  period.
          See also the _p_h_a_s_e parameter.

     _s_t_a_b_i_l_i_t_y - _f_l_o_a_t

          Stability factor in -dB.

     _p_h_a_s_e - _f_l_o_a_t
          This is a number between 0 and 1.  Value  0  means  the
          location  of  the window center will be the same as the
          epoch pulse location.  A value of 0.75,  (the  default)
          causes  the  window to be centered in the middle of the
          "closed glottis" region, assuming an "open quotient" of
          0.5.

EEEESSSSPPPPSSSS CCCCOOOOMMMMMMMMOOOONNNN
     No ESPS common parameter processing is supported.

EEEESSSSPPPPSSSS HHHHEEEEAAAADDDDEEEERRRRSSSS
     The usual _r_e_c_o_r_d__f_r_e_q, _s_t_a_r_t__t_i_m_e header items, and all sup-
     ported ESPS parameters are stored as generic header items.

     The bandwidth of _i_n__d_a_t_a, usually one half of  its  sampling
     frequency, is recorded as a generic header item _b_a_n_d_w_i_d_t_h of
     DOUBLE type.

FFFFUUUUTTTTUUUURRRREEEE CCCCHHHHAAAANNNNGGGGEEEESSSS
EEEEXXXXAAAAMMMMPPPPLLLLEEEESSSS
     The manual page for _e_p_o_c_h_s includes an  example  of  glottal
     epoch  detection  such  as  might  be performed before using
     _p_s__a_n_a.  Assuming  one  has  a  file  "spch.pe"   containing
     impulses at the estimated instants of glottal closure, and a
     speech file called "spch.sd",  the  following  command  will
     produce  the  files  "spch.f0ps" and "spch.rc".  The ".f0ps"
     file has F0 estimates obtained by inverting the  inter-pulse
     intervals  from  the  ".pe"  file  and RMS computed from the
     epoch-synchronous windows of preemphasized data used by  the
     LPC analysis.

     ps_ana -f spch.f0ps spch.sd spch.pe spch.rc

EEEERRRRRRRROOOORRRRSSSS AAAANNNNDDDD DDDDIIIIAAAAGGGGNNNNOOOOSSSSTTTTIIIICCCCSSSS
BBBBUUUUGGGGSSSS
     None known.

RRRREEEEFFFFEEEERRRREEEENNNNCCCCEEEESSSS
     Talkin, D. and Rowley, J., "Pitch-Synchronous  analysis  and



        Copyright Entropic Research Laboratory, Inc. 4G4        4






PS_ANA(1-ESPS)                              ENTROPIC RESEARCH LAB



     synthesis for TTS systems," _P_r_o_c_e_e_d_i_n_g_s _o_f _t_h_e _E_S_C_A _W_o_r_k_s_h_o_p
     _o_n  _S_p_e_e_c_h  _S_y_n_t_h_e_s_i_s,  C.  Benoit,  Ed.,   Imprimerie   des
     Ecureuils, Gieres, France, 1990.

SSSSEEEEEEEE AAAALLLLSSSSOOOO
     lp_syn(1-ESPS), formant(1-ESPS), get_f0(1-ESPS),
     epochs(1-ESPS), get_resid(1-ESPS), FEA(5-ESPS),
     FEA_SD(5-ESPS)

AAAAUUUUTTTTHHHHOOOORRRR
     David Talkin, Derek Lin









































        Copyright Entropic Research Laboratory, Inc. 5G5        5






