2 * Copyright (c) 2012 Stefano Sabatini
4 * This file is part of FFmpeg.
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23 * flite voice synth source
26 #include <flite/flite.h>
27 #include "libavutil/channel_layout.h"
28 #include "libavutil/file.h"
29 #include "libavutil/opt.h"
35 typedef struct FliteContext {
41 int16_t *wave_samples;
45 struct voice_entry *voice_entry;
47 int frame_nb_samples; ///< number of samples per frame
50 #define OFFSET(x) offsetof(FliteContext, x)
51 #define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
53 static const AVOption flite_options[] = {
54 { "list_voices", "list voices and exit", OFFSET(list_voices), AV_OPT_TYPE_BOOL, {.i64=0}, 0, 1, FLAGS },
55 { "nb_samples", "set number of samples per frame", OFFSET(frame_nb_samples), AV_OPT_TYPE_INT, {.i64=512}, 0, INT_MAX, FLAGS },
56 { "n", "set number of samples per frame", OFFSET(frame_nb_samples), AV_OPT_TYPE_INT, {.i64=512}, 0, INT_MAX, FLAGS },
57 { "text", "set text to speak", OFFSET(text), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, FLAGS },
58 { "textfile", "set filename of the text to speak", OFFSET(textfile), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, FLAGS },
59 { "v", "set voice", OFFSET(voice_str), AV_OPT_TYPE_STRING, {.str="kal"}, 0, 0, FLAGS },
60 { "voice", "set voice", OFFSET(voice_str), AV_OPT_TYPE_STRING, {.str="kal"}, 0, 0, FLAGS },
64 AVFILTER_DEFINE_CLASS(flite);
66 static volatile int flite_inited = 0;
68 /* declare functions for all the supported voices */
69 #define DECLARE_REGISTER_VOICE_FN(name) \
70 cst_voice *register_cmu_us_## name(const char *); \
71 void unregister_cmu_us_## name(cst_voice *);
72 DECLARE_REGISTER_VOICE_FN(awb);
73 DECLARE_REGISTER_VOICE_FN(kal);
74 DECLARE_REGISTER_VOICE_FN(kal16);
75 DECLARE_REGISTER_VOICE_FN(rms);
76 DECLARE_REGISTER_VOICE_FN(slt);
80 cst_voice * (*register_fn)(const char *);
81 void (*unregister_fn)(cst_voice *);
86 #define MAKE_VOICE_STRUCTURE(voice_name) { \
87 .name = #voice_name, \
88 .register_fn = register_cmu_us_ ## voice_name, \
89 .unregister_fn = unregister_cmu_us_ ## voice_name, \
91 static struct voice_entry voice_entries[] = {
92 MAKE_VOICE_STRUCTURE(awb),
93 MAKE_VOICE_STRUCTURE(kal),
94 MAKE_VOICE_STRUCTURE(kal16),
95 MAKE_VOICE_STRUCTURE(rms),
96 MAKE_VOICE_STRUCTURE(slt),
99 static void list_voices(void *log_ctx, const char *sep)
101 int i, n = FF_ARRAY_ELEMS(voice_entries);
102 for (i = 0; i < n; i++)
103 av_log(log_ctx, AV_LOG_INFO, "%s%s",
104 voice_entries[i].name, i < (n-1) ? sep : "\n");
107 static int select_voice(struct voice_entry **entry_ret, const char *voice_name, void *log_ctx)
111 for (i = 0; i < FF_ARRAY_ELEMS(voice_entries); i++) {
112 struct voice_entry *entry = &voice_entries[i];
113 if (!strcmp(entry->name, voice_name)) {
115 entry->voice = entry->register_fn(NULL);
117 av_log(log_ctx, AV_LOG_ERROR,
118 "Could not register voice '%s'\n", voice_name);
119 return AVERROR_UNKNOWN;
121 entry->usage_count++;
127 av_log(log_ctx, AV_LOG_ERROR, "Could not find voice '%s'\n", voice_name);
128 av_log(log_ctx, AV_LOG_INFO, "Choose between the voices: ");
129 list_voices(log_ctx, ", ");
131 return AVERROR(EINVAL);
134 static av_cold int init(AVFilterContext *ctx)
136 FliteContext *flite = ctx->priv;
139 if (flite->list_voices) {
140 list_voices(ctx, "\n");
145 if (flite_init() < 0) {
146 av_log(ctx, AV_LOG_ERROR, "flite initialization failed\n");
147 return AVERROR_UNKNOWN;
152 if ((ret = select_voice(&flite->voice_entry, flite->voice_str, ctx)) < 0)
154 flite->voice = flite->voice_entry->voice;
156 if (flite->textfile && flite->text) {
157 av_log(ctx, AV_LOG_ERROR,
158 "Both text and textfile options set: only one must be specified\n");
159 return AVERROR(EINVAL);
162 if (flite->textfile) {
166 if ((ret = av_file_map(flite->textfile, &textbuf, &textbuf_size, 0, ctx)) < 0) {
167 av_log(ctx, AV_LOG_ERROR,
168 "The text file '%s' could not be read: %s\n",
169 flite->textfile, av_err2str(ret));
173 if (!(flite->text = av_malloc(textbuf_size+1))) {
174 av_file_unmap(textbuf, textbuf_size);
175 return AVERROR(ENOMEM);
177 memcpy(flite->text, textbuf, textbuf_size);
178 flite->text[textbuf_size] = 0;
179 av_file_unmap(textbuf, textbuf_size);
183 av_log(ctx, AV_LOG_ERROR,
184 "No speech text specified, specify the 'text' or 'textfile' option\n");
185 return AVERROR(EINVAL);
188 /* synth all the file data in block */
189 flite->wave = flite_text_to_wave(flite->text, flite->voice);
190 flite->wave_samples = flite->wave->samples;
191 flite->wave_nb_samples = flite->wave->num_samples;
195 static av_cold void uninit(AVFilterContext *ctx)
197 FliteContext *flite = ctx->priv;
199 if (!--flite->voice_entry->usage_count)
200 flite->voice_entry->unregister_fn(flite->voice);
202 flite->voice_entry = NULL;
203 delete_wave(flite->wave);
207 static int query_formats(AVFilterContext *ctx)
209 FliteContext *flite = ctx->priv;
212 AVFilterChannelLayouts *chlayouts = NULL;
213 int64_t chlayout = av_get_default_channel_layout(flite->wave->num_channels);
214 AVFilterFormats *sample_formats = NULL;
215 AVFilterFormats *sample_rates = NULL;
217 if ((ret = ff_add_channel_layout (&chlayouts , chlayout )) < 0 ||
218 (ret = ff_set_common_channel_layouts (ctx , chlayouts )) < 0 ||
219 (ret = ff_add_format (&sample_formats, AV_SAMPLE_FMT_S16 )) < 0 ||
220 (ret = ff_set_common_formats (ctx , sample_formats )) < 0 ||
221 (ret = ff_add_format (&sample_rates , flite->wave->sample_rate)) < 0 ||
222 (ret = ff_set_common_samplerates (ctx , sample_rates )) < 0)
228 static int config_props(AVFilterLink *outlink)
230 AVFilterContext *ctx = outlink->src;
231 FliteContext *flite = ctx->priv;
233 outlink->sample_rate = flite->wave->sample_rate;
234 outlink->time_base = (AVRational){1, flite->wave->sample_rate};
236 av_log(ctx, AV_LOG_VERBOSE, "voice:%s fmt:%s sample_rate:%d\n",
238 av_get_sample_fmt_name(outlink->format), outlink->sample_rate);
242 static int request_frame(AVFilterLink *outlink)
245 FliteContext *flite = outlink->src->priv;
246 int nb_samples = FFMIN(flite->wave_nb_samples, flite->frame_nb_samples);
251 samplesref = ff_get_audio_buffer(outlink, nb_samples);
253 return AVERROR(ENOMEM);
255 memcpy(samplesref->data[0], flite->wave_samples,
256 nb_samples * flite->wave->num_channels * 2);
257 samplesref->pts = flite->pts;
258 samplesref->pkt_pos = -1;
259 samplesref->sample_rate = flite->wave->sample_rate;
260 flite->pts += nb_samples;
261 flite->wave_samples += nb_samples * flite->wave->num_channels;
262 flite->wave_nb_samples -= nb_samples;
264 return ff_filter_frame(outlink, samplesref);
267 static const AVFilterPad flite_outputs[] = {
270 .type = AVMEDIA_TYPE_AUDIO,
271 .config_props = config_props,
272 .request_frame = request_frame,
277 AVFilter ff_asrc_flite = {
279 .description = NULL_IF_CONFIG_SMALL("Synthesize voice from text using libflite."),
280 .query_formats = query_formats,
283 .priv_size = sizeof(FliteContext),
285 .outputs = flite_outputs,
286 .priv_class = &flite_class,