rockbox/apps/talk.h

184 lines
7.3 KiB
C
Raw Normal View History

/***************************************************************************
* __________ __ ___.
* Open \______ \ ____ ____ | | _\_ |__ _______ ___
* Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ /
* Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < <
* Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \
* \/ \/ \/ \/ \/
* $Id$
*
* Copyright (C) 2004 Jörg Hohensohn
*
* This module collects the Talkbox and voice UI functions.
* (Talkbox reads directory names from mp3 clips called thumbnails,
* the voice UI lets menus and screens "talk" from a voicefont in memory.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
* of the License, or (at your option) any later version.
*
* This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
* KIND, either express or implied.
*
****************************************************************************/
#ifndef __TALK_H__
#define __TALK_H__
#include <stdbool.h>
#include <inttypes.h>
#include "time.h"
#define VOICE_VERSION 400 /* 4.00 - if you change this, change it in voicefont too */
enum {
/* See array "unit_voiced" in talk.c function "talk_value" */
UNIT_INT = 1, /* plain number */
UNIT_SIGNED, /* number with mandatory sign (even if positive) */
UNIT_MS, /* milliseconds */
UNIT_SEC, /* seconds */
UNIT_MIN, /* minutes */
UNIT_HOUR, /* hours */
UNIT_KHZ, /* kHz */
UNIT_DB, /* dB, mandatory sign */
UNIT_PERCENT, /* % */
UNIT_MAH, /* milliAmp hours */
UNIT_PIXEL, /* pixels */
UNIT_PER_SEC, /* per second */
UNIT_HERTZ, /* hertz */
UNIT_MB, /* Megabytes */
UNIT_KBIT, /* kilobits per sec */
UNIT_PM_TICK, /* peak meter units per tick */
UNIT_TIME, /* time duration/interval in seconds, says hours,mins,secs */
UNIT_DATEYEAR,/* for 1999 say nineteen ninety nine */
UNIT_LAST /* END MARKER */
};
#define UNIT_SHIFT (32-5) /* this many bits left from UNIT_xx enum */
#define DECIMAL_SHIFT (32 - 8)
/* make a "talkable" ID from number + unit
unit is upper 4 bits, number the remaining (in regular 2's complement) */
#define TALK_ID(n,u) (((long)(u))<<UNIT_SHIFT | ((n) & ~(-1L<<UNIT_SHIFT)))
/* make a "talkable" ID from a decimal number + unit, the decimal number
is represented like x*10^d where d is the number of decimal digits */
#define TALK_ID_DECIMAL(n,d,u) (((long)(u))<<UNIT_SHIFT |\
((long)(d))<<DECIMAL_SHIFT |\
((n) & ~(-1L<<DECIMAL_SHIFT)))
/* convenience macro to have both virtual pointer and ID as arguments */
#define STR(id) ID2P(id), id
/* Policy values for how hard to try to keep the talk/voice buffers.
* Affects how genereous talk.c is when it's asked for memory in
* shrink_callbacks().
*
* I.e. setting the policy to TALK_BUFFER_LOOSE, it will happily give its
* entire bufer away if asked for, e.g. due to a another module
* calling core_alloc_maximum(), TALK_BUFFER_HOLD on the other hand will
* make it keep the buffers so that a call to core_alloc_maximum() does not
* stop the speech-interface.
*/
enum talk_buffer_policies {
TALK_BUFFER_DEFAULT,
TALK_BUFFER_LOOSE,
TALK_BUFFER_HOLD,
};
/* This sets the actual policy. Call this before core_alloc_maximum() to
* get the desired outcome */
void talk_buffer_set_policy(int policy);
/* publish these strings, so they're stored only once (better than #define) */
extern const char* const dir_thumbnail_name; /* "_dirname.talk" */
extern const char* const file_thumbnail_ext; /* ".talk" for file voicing */
void talk_init(void);
int talk_get_bufsize(void); /* get the loaded voice file size */
size_t talkbuf_init(char* bufstart);
bool is_voice_queued(void); /* Are there more voice clips to be spoken? */
int talk_id(int32_t id, bool enqueue); /* play a voice ID from voicefont */
/* play a thumbnail from file */
int talk_file(const char *root, const char *dir, const char *file,
const char *ext, const long *prefix_ids, bool enqueue);
/* play file's thumbnail or spell name */
int talk_file_or_spell(const char *dirname, const char* filename,
const long *prefix_ids, bool enqueue);
#if CONFIG_CODEC == SWCODEC
/* play dir's thumbnail or spell name */
int talk_dir_or_spell(const char* filename,
const long *prefix_ids, bool enqueue);
#endif
/* play thumbnails for each components of full path, or spell */
int talk_fullpath(const char* path, bool enqueue);
int talk_number(long n, bool enqueue); /* say a number */
int talk_value(long n, int unit, bool enqueue); /* say a numeric value */
int talk_value_decimal(long n, int unit, int decimals, bool enqueue);
int talk_spell(const char* spell, bool enqueue); /* spell a string */
void talk_setting(const void *global_settings_variable); /* read a setting */
void talk_disable(bool disable); /* temporarily disable (or re-enable) talking (temporarily, not persisted) */
void talk_force_shutup(void); /* kill voice unconditionally */
void talk_shutup(void); /* Interrupt voice, as when enqueue is false */
/* helper function for speaking fractional numbers */
void talk_fractional(char *tbuf, int value, int unit);
#if CONFIG_RTC
void talk_time(const struct tm *tm, bool enqueue);
void talk_date(const struct tm *tm, bool enqueue);
#endif /* CONFIG_RTC */
/* speaks hr, min, sec, ms; unit_idx is lowest or base unit of the time value */
int talk_time_intervals(long time, int unit_idx, bool enqueue);
/* This (otherwise invalid) ID signals the end of the array. */
#define TALK_FINAL_ID LANG_LAST_INDEX_IN_ARRAY
/* Enqueue next utterance even if enqueue parameter is false: don't
interrupt the current utterance. */
void talk_force_enqueue_next(void);
/* speaks one or more IDs (from an array)). */
int talk_idarray(const long *idarray, bool enqueue);
/* This makes an initializer for the array of IDs and takes care to
put the final sentinel element at the end. */
#define TALK_IDARRAY(ids...) ((long[]){ids,TALK_FINAL_ID})
/* And this handy macro makes it look like a variadic function. */
#define talk_ids(enqueue, ids...) talk_idarray(TALK_IDARRAY(ids), enqueue)
/* This version talks only if talking menus are enabled, and does not
enqueue the initial id. */
#define cond_talk_ids(ids...) do { \
if (global_settings.talk_menu) \
talk_ids(false, ids); \
} while(0)
/* And a version that takes the array parameter... */
#define cond_talk_idarray(idarray) do { \
if (global_settings.talk_menu \
talk_idarray(idarray, false); \
} while(0)
/* Convenience macro to conditionally speak something and not have
it interrupted. */
#define cond_talk_ids_fq(ids...) do { \
if (global_settings.talk_menu) { \
talk_ids(false, ids); \
talk_force_enqueue_next(); \
} \
}while(0)
struct talk_debug_data {
char voicefile[32];
long memory_allocated, memory_used;
int num_clips, num_empty_clips;
int min_clipsize, avg_clipsize, max_clipsize;
int cached_clips;
int cache_hits;
int cache_misses;
};
bool talk_get_debug_data(struct talk_debug_data *data);
#endif /* __TALK_H__ */