2 * This file is part of FFmpeg.
4 * FFmpeg is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU Lesser General Public
6 * License as published by the Free Software Foundation; either
7 * version 2.1 of the License, or (at your option) any later version.
9 * FFmpeg is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * Lesser General Public License for more details.
14 * You should have received a copy of the GNU Lesser General Public
15 * License along with FFmpeg; if not, write to the Free Software
16 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19 #include <rubberband/rubberband-c.h>
21 #include "libavutil/channel_layout.h"
22 #include "libavutil/common.h"
23 #include "libavutil/opt.h"
31 typedef struct RubberBandContext {
36 int transients, detector, phase, window,
37 smoothing, formant, opitch, channels;
38 int64_t nb_samples_out;
39 int64_t nb_samples_in;
44 #define OFFSET(x) offsetof(RubberBandContext, x)
45 #define A AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
47 static const AVOption rubberband_options[] = {
48 { "tempo", "set tempo scale factor", OFFSET(tempo), AV_OPT_TYPE_DOUBLE, {.dbl=1}, 0.01, 100, A },
49 { "pitch", "set pitch scale factor", OFFSET(pitch), AV_OPT_TYPE_DOUBLE, {.dbl=1}, 0.01, 100, A },
50 { "transients", "set transients", OFFSET(transients), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX, A, "transients" },
51 { "crisp", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionTransientsCrisp}, 0, 0, A, "transients" },
52 { "mixed", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionTransientsMixed}, 0, 0, A, "transients" },
53 { "smooth", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionTransientsSmooth}, 0, 0, A, "transients" },
54 { "detector", "set detector", OFFSET(detector), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX, A, "detector" },
55 { "compound", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionDetectorCompound}, 0, 0, A, "detector" },
56 { "percussive", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionDetectorPercussive}, 0, 0, A, "detector" },
57 { "soft", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionDetectorSoft}, 0, 0, A, "detector" },
58 { "phase", "set phase", OFFSET(phase), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX, A, "phase" },
59 { "laminar", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionPhaseLaminar}, 0, 0, A, "phase" },
60 { "independent", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionPhaseIndependent}, 0, 0, A, "phase" },
61 { "window", "set window", OFFSET(window), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX, A, "window" },
62 { "standard", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionWindowStandard}, 0, 0, A, "window" },
63 { "short", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionWindowShort}, 0, 0, A, "window" },
64 { "long", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionWindowLong}, 0, 0, A, "window" },
65 { "smoothing", "set smoothing", OFFSET(smoothing), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX, A, "smoothing" },
66 { "off", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionSmoothingOff}, 0, 0, A, "smoothing" },
67 { "on", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionSmoothingOn}, 0, 0, A, "smoothing" },
68 { "formant", "set formant", OFFSET(formant), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX, A, "formant" },
69 { "shifted", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionFormantShifted}, 0, 0, A, "formant" },
70 { "preserved", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionFormantPreserved}, 0, 0, A, "formant" },
71 { "pitchq", "set pitch quality", OFFSET(opitch), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX, A, "pitch" },
72 { "quality", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionPitchHighQuality}, 0, 0, A, "pitch" },
73 { "speed", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionPitchHighSpeed}, 0, 0, A, "pitch" },
74 { "consistency", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionPitchHighConsistency}, 0, 0, A, "pitch" },
75 { "channels", "set channels", OFFSET(channels), AV_OPT_TYPE_INT, {.i64=0}, 0, INT_MAX, A, "channels" },
76 { "apart", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionChannelsApart}, 0, 0, A, "channels" },
77 { "together", 0, 0, AV_OPT_TYPE_CONST, {.i64=RubberBandOptionChannelsTogether}, 0, 0, A, "channels" },
81 AVFILTER_DEFINE_CLASS(rubberband);
83 static av_cold void uninit(AVFilterContext *ctx)
85 RubberBandContext *s = ctx->priv;
88 rubberband_delete(s->rbs);
91 static int query_formats(AVFilterContext *ctx)
93 AVFilterFormats *formats = NULL;
94 AVFilterChannelLayouts *layouts = NULL;
95 static const enum AVSampleFormat sample_fmts[] = {
101 layouts = ff_all_channel_counts();
103 return AVERROR(ENOMEM);
104 ret = ff_set_common_channel_layouts(ctx, layouts);
108 formats = ff_make_format_list(sample_fmts);
110 return AVERROR(ENOMEM);
111 ret = ff_set_common_formats(ctx, formats);
115 formats = ff_all_samplerates();
117 return AVERROR(ENOMEM);
118 return ff_set_common_samplerates(ctx, formats);
121 static int filter_frame(AVFilterLink *inlink, AVFrame *in)
123 RubberBandContext *s = inlink->dst->priv;
124 AVFilterLink *outlink = inlink->dst->outputs[0];
126 int ret = 0, nb_samples;
128 if (s->first_pts == AV_NOPTS_VALUE)
129 s->first_pts = in->pts;
131 rubberband_process(s->rbs, (const float *const *)in->data, in->nb_samples, ff_outlink_get_status(inlink));
132 s->nb_samples_in += in->nb_samples;
134 nb_samples = rubberband_available(s->rbs);
135 if (nb_samples > 0) {
136 out = ff_get_audio_buffer(outlink, nb_samples);
139 return AVERROR(ENOMEM);
141 out->pts = s->first_pts + av_rescale_q(s->nb_samples_out,
142 (AVRational){ 1, outlink->sample_rate },
144 nb_samples = rubberband_retrieve(s->rbs, (float *const *)out->data, nb_samples);
145 out->nb_samples = nb_samples;
146 ret = ff_filter_frame(outlink, out);
147 s->nb_samples_out += nb_samples;
154 static int config_input(AVFilterLink *inlink)
156 AVFilterContext *ctx = inlink->dst;
157 RubberBandContext *s = ctx->priv;
158 int opts = s->transients|s->detector|s->phase|s->window|
159 s->smoothing|s->formant|s->opitch|s->channels|
160 RubberBandOptionProcessRealTime;
163 rubberband_delete(s->rbs);
164 s->rbs = rubberband_new(inlink->sample_rate, inlink->channels, opts, 1. / s->tempo, s->pitch);
166 return AVERROR(ENOMEM);
168 s->nb_samples = rubberband_get_samples_required(s->rbs);
169 s->first_pts = AV_NOPTS_VALUE;
174 static int activate(AVFilterContext *ctx)
176 AVFilterLink *inlink = ctx->inputs[0];
177 AVFilterLink *outlink = ctx->outputs[0];
178 RubberBandContext *s = ctx->priv;
182 FF_FILTER_FORWARD_STATUS_BACK(outlink, inlink);
184 ret = ff_inlink_consume_samples(inlink, s->nb_samples, s->nb_samples, &in);
188 return filter_frame(inlink, in);
190 FF_FILTER_FORWARD_STATUS(inlink, outlink);
191 FF_FILTER_FORWARD_WANTED(outlink, inlink);
193 return FFERROR_NOT_READY;
196 static int process_command(AVFilterContext *ctx, const char *cmd, const char *args,
197 char *res, int res_len, int flags)
199 RubberBandContext *s = ctx->priv;
201 if (!strcmp(cmd, "tempo")) {
204 sscanf(args, "%lf", &arg);
205 if (arg < 0.01 || arg > 100) {
206 av_log(ctx, AV_LOG_ERROR,
207 "Tempo scale factor '%f' out of range\n", arg);
208 return AVERROR(EINVAL);
210 rubberband_set_time_ratio(s->rbs, 1. / arg);
213 if (!strcmp(cmd, "pitch")) {
216 sscanf(args, "%lf", &arg);
217 if (arg < 0.01 || arg > 100) {
218 av_log(ctx, AV_LOG_ERROR,
219 "Pitch scale factor '%f' out of range\n", arg);
220 return AVERROR(EINVAL);
222 rubberband_set_pitch_scale(s->rbs, arg);
228 static const AVFilterPad rubberband_inputs[] = {
231 .type = AVMEDIA_TYPE_AUDIO,
232 .config_props = config_input,
237 static const AVFilterPad rubberband_outputs[] = {
240 .type = AVMEDIA_TYPE_AUDIO,
245 AVFilter ff_af_rubberband = {
246 .name = "rubberband",
247 .description = NULL_IF_CONFIG_SMALL("Apply time-stretching and pitch-shifting."),
248 .query_formats = query_formats,
249 .priv_size = sizeof(RubberBandContext),
250 .priv_class = &rubberband_class,
252 .activate = activate,
253 .inputs = rubberband_inputs,
254 .outputs = rubberband_outputs,
255 .process_command = process_command,