<sect1 id='tilt-analysis-manual'> <title><command>tilt_analysis</command> <emphasis>Produce tilt descriptions from F0 contours</emphasis></title> <toc depth='1'></toc> <para> </para> <sect2> <title>Synopsis</title> <para> </para> <!-- /amd/projects/festival/versions/v_mpiro/speech_tools_linux/bin/tilt_analysis -sgml_synopsis --> <para> <cmdsynopsis><command>tilt_analysis</command>[input f0 file] -e [input event label file] -o [output file][options]Summary: produce rfc file from events and f0 contour<arg>-h </arg> <arg>-itype <replaceable>string</replaceable></arg> <arg>-ctype <replaceable>string</replaceable></arg> <arg>-s <replaceable>float</replaceable></arg> <arg>-c <replaceable>string</replaceable></arg> <arg>-start <replaceable>float</replaceable></arg> <arg>-end <replaceable>float</replaceable></arg> <arg>-from <replaceable>int</replaceable></arg> <arg>-to <replaceable>int</replaceable></arg> <arg>-event_names <replaceable>string</replaceable></arg> <arg>-sil_names <replaceable>string</replaceable></arg> <arg>-e <replaceable>ifile</replaceable></arg> <arg>-o <replaceable>ofile</replaceable></arg> <arg>-otype <replaceable>string</replaceable></arg> <arg>-limit <replaceable>float</replaceable></arg> <arg>-range <replaceable>float</replaceable></arg> <arg>-smooth </arg> <arg>-w1 <replaceable>float</replaceable></arg> <arg>-w2 <replaceable>float</replaceable></arg> <arg>-sf0 <replaceable>ofile</replaceable></arg> <arg>-rfc </arg> </cmdsynopsis> </para> <!-- DONE /amd/projects/festival/versions/v_mpiro/speech_tools_linux/bin/tilt_analysis -sgml_synopsis --> <para> tilt_analysis produces a Tilt or RFC analysis of a F0 contour, given a set label file containing a set of approximate intonational event boundaries. A detailed description of the Tilt intonation model can be found in the <link linkend="tilt-overview">Tilt model overview</link> section. </para> </sect2> <sect2> <title>OPTIONS</title> <para> </para> <!-- /amd/projects/festival/versions/v_mpiro/speech_tools_linux/bin/tilt_analysis -sgml_options --> <para> <variablelist> <varlistentry><term>-h</term> <LISTITEM><PARA> Options help </PARA></LISTITEM> </varlistentry> <varlistentry><term>-itype</term> <LISTITEM><PARA> <replaceable>string</replaceable> Input file type (optional). If no type is specified type is automatically derived from file's header. Supported types are: none, esps, est, est_binary, htk, htk_fbank, htk_mfcc, htk_user, htk_discrete, ssff, xmg, xgraph, ema, ema_swapped, ascii </PARA></LISTITEM> </varlistentry> <varlistentry><term>-ctype</term> <LISTITEM><PARA> <replaceable>string</replaceable> Contour type: F0, track </PARA></LISTITEM> </varlistentry> <varlistentry><term>-s</term> <LISTITEM><PARA> <replaceable>float</replaceable> Frame spacing of input in seconds, for unheadered input file </PARA></LISTITEM> </varlistentry> <varlistentry><term>-c</term> <LISTITEM><PARA> <replaceable>string</replaceable> Select a subset of channels (starts from 0). Tracks can have multiple channels. This option specifies a list of numbers, refering to the channel numbers which are to be used for for processing. </PARA></LISTITEM> </varlistentry> <varlistentry><term>-start</term> <LISTITEM><PARA> <replaceable>float</replaceable> Extract track starting at this time, specified in seconds </PARA></LISTITEM> </varlistentry> <varlistentry><term>-end</term> <LISTITEM><PARA> <replaceable>float</replaceable> Extract track ending at this time, specified in seconds </PARA></LISTITEM> </varlistentry> <varlistentry><term>-from</term> <LISTITEM><PARA> <replaceable>int</replaceable> Extract track starting at this frame position </PARA></LISTITEM> </varlistentry> <varlistentry><term>-to</term> <LISTITEM><PARA> <replaceable>int</replaceable> Extract track ending at this frame position </PARA></LISTITEM> </varlistentry> <varlistentry><term>-event_names</term> <LISTITEM><PARA> <replaceable>string</replaceable> List of labels to be classed as events. Lists are specified as quoted strings with spaces separating each item, e.g.: "a b c d" </PARA></LISTITEM> </varlistentry> <varlistentry><term>-sil_names</term> <LISTITEM><PARA> <replaceable>string</replaceable> List of labels to be classed as silence Lists are specified as quoted strings with spaces separating each item, e.g.: "pau sil #" </PARA></LISTITEM> </varlistentry> <varlistentry><term>-e</term> <LISTITEM><PARA> <replaceable>ifile</replaceable> Input event label file. This file contains the list of events to be parameterized, each with its approximate start and stop time marked. This file also contains silencesn which are used to decide where to insert and stop phrases </PARA></LISTITEM> </varlistentry> <varlistentry><term>-o</term> <LISTITEM><PARA> <replaceable>ofile</replaceable> Output label file </PARA></LISTITEM> </varlistentry> <varlistentry><term>-otype</term> <LISTITEM><PARA> <replaceable>string</replaceable> File type of output file </PARA></LISTITEM> </varlistentry> <varlistentry><term>-limit</term> <LISTITEM><PARA> <replaceable>float</replaceable> start and stop limit in seconds. The rfc matching algorithm defines a search region within which it tries all possible rise and fall shapes. This option specifies how much before the input label start time and how much after the input label end time the search region should be. Typical value, 0.1 </PARA></LISTITEM> </varlistentry> <varlistentry><term>-range</term> <LISTITEM><PARA> <replaceable>float</replaceable> Range of RFC search region. In addition to the limit, the range defines the limits of the rfc matching search region as a percentage of the overal input label duration. Typical value, 0.25 (the search region is the first and last 25% of the label) </PARA></LISTITEM> </varlistentry> <varlistentry><term>-smooth</term> <LISTITEM><PARA> Smooth and Interpolate input F0 contour. rfc matching can only operate on smooth fully interpolated contours. This option must be used if the contour hasn't already been smoothed and interpolated </PARA></LISTITEM> </varlistentry> <varlistentry><term>-w1</term> <LISTITEM><PARA> <replaceable>float</replaceable> length in seconds of smoothing window prior to interpolation. Default value 0.05 </PARA></LISTITEM> </varlistentry> <varlistentry><term>-w2</term> <LISTITEM><PARA> <replaceable>float</replaceable> length in seconds of smoothing window after to interpolation. Default value 0.05 </PARA></LISTITEM> </varlistentry> <varlistentry><term>-sf0</term> <LISTITEM><PARA> <replaceable>ofile</replaceable> Save f0 contour that results from smoothing </PARA></LISTITEM> </varlistentry> <varlistentry><term>-rfc</term> <LISTITEM><PARA> Save as RFC parameters instead of tilt </PARA></LISTITEM> </varlistentry> </variablelist> </para> <!-- DONE /amd/projects/festival/versions/v_mpiro/speech_tools_linux/bin/tilt_analysis -sgml_options --> </sect2> <sect2> <title>Input Intonation Files</title> <para> A label file containing approximate intonational event bounaries must be given as input. A typical file in xlabel format is shown below: </para> <para> <screen> 0.290 146 sil 0.480 146 c 0.620 146 a 0.760 146 c 0.960 146 a 1.480 146 c 1.680 146 a 1.790 146 sil </screen> </para> <para> The set of intonational events can be given on the command line with the -event_names option. The default set is "a rb arb m mrb" and so the above example would not need the -event_names option. The label "c" (connection) is to separate events, in effect giving each event a start time as well as a end time. The silence labels are important also: they specify where phrases should start and end. </para> </sect2> <sect2> <title>Input F0 Files</title> <para> tilt_analysis can operate on all the F0 file types supported by the EST library. Tilt analysis can only operate on smooth and continuous F0 contours.(i.e. F0 values must be defined during unvoiced regons). If the input contour is not in this format, use the -smooth option. The -w1 and -w2 options can be used to control the amount of smoothing. The smoothed version of the input contour can be examined by saving it using the -sf0 option. </para> </sect2> <sect2> <title>Output Intonation Files</title> <para> The output will be a label file containing the tilt parameters for the events in feature format. An example, in xlabel format, is shown below: </para> <para> <screen> intonation_style tilt # 0.29 26 phrase_start ; ev.f0 115.234 ; time 0.29 ; 0.53 26 a ; int_event 1 ; ev.f0 118.171 ; time 0.53 ; tilt.amp 21.8602 ; tilt.dur 0.26 ; tilt.tilt -0.163727 ; 0.77 26 a ; int_event 1 ; ev.f0 112.694 ; time 0.77 ; tilt.amp 27.0315 ; tilt.dur 0.32 ; tilt.tilt -0.446791 ; 1.53 26 a ; int_event 1 ; ev.f0 100.83 ; time 1.53 ; tilt.amp 7.507 ; tilt.dur 0.22 ; tilt.tilt -0.296317 ; 1.79 26 phrase_end ; ev.f0 92.9785 ; time 1.79 ; </screen> </para> <para> The -rfc option will make a file containing the RFC parameters instead: </para> <para> <screen> intonation_style rfc # 0.29 26 phrase_start ; ev.f0 115.234 ; time 0.29 ; 0.53 26 a ; ev.f0 118.171 ; rfc.rise_amp 8.19178 ; rfc.rise_dur 0.12 ; rfc.fall_amp -13.6684 ; rfc.fall_dur 0.14 ; time 0.53 ; 0.77 26 a ; ev.f0 112.694 ; rfc.rise_amp 6.50673 ; rfc.rise_dur 0.1 ; rfc.fall_amp -20.5248 ; rfc.fall_dur 0.22 ; time 0.77 ; 1.53 26 a ; ev.f0 100.83 ; rfc.rise_amp 1.55832 ; rfc.rise_dur 0.11 ; rfc.fall_amp -6.09238 ; rfc.fall_dur 0.11 ; time 1.53 ; 1.79 26 phrase_end ; ev.f0 92.9785 ; time 1.79 ; </screen> </para> <para> The feature in the header, "intonation_style tilt" or "intonation_style rfc" is needed for the tilt_synthesis program to work. </para> </sect2> </sect1>