static void noise_gate_update(void *data, obs_data_t *s)
{
struct noise_gate_data *ng = data;
float open_threshold_db;
float close_threshold_db;
float sample_rate;
int attack_time_ms;
int hold_time_ms;
int release_time_ms;
open_threshold_db = (float)obs_data_get_double(s, S_OPEN_THRESHOLD);
close_threshold_db = (float)obs_data_get_double(s, S_CLOSE_THRESHOLD);
attack_time_ms = (int)obs_data_get_int(s, S_ATTACK_TIME);
hold_time_ms = (int)obs_data_get_int(s, S_HOLD_TIME);
release_time_ms = (int)obs_data_get_int(s, S_RELEASE_TIME);
sample_rate = (float)audio_output_get_sample_rate(obs_get_audio());
ng->sample_rate_i = 1.0f / sample_rate;
ng->channels = audio_output_get_channels(obs_get_audio());
ng->open_threshold = db_to_mul(open_threshold_db);
ng->close_threshold = db_to_mul(close_threshold_db);
ng->attack_rate = 1.0f / (ms_to_secf(attack_time_ms) * sample_rate);
ng->release_rate = 1.0f / (ms_to_secf(release_time_ms) * sample_rate);
const float threshold_diff = ng->open_threshold - ng->close_threshold;
const float min_decay_period = (1.0f / 75.0f) * sample_rate;
ng->decay_rate = threshold_diff / min_decay_period;
ng->hold_time = ms_to_secf(hold_time_ms);
ng->is_open = false;
ng->attenuation = 0.0f;
ng->level = 0.0f;
ng->held_time = 0.0f;
}
static void expander_update(void *data, obs_data_t *s)
{
struct expander_data *cd = data;
const char *presets = obs_data_get_string(s, S_PRESETS);
if (strcmp(presets, "expander") == 0 && cd->is_gate) {
obs_data_clear(s);
obs_data_set_string(s, S_PRESETS, "expander");
expander_defaults(s);
cd->is_gate = false;
}
if (strcmp(presets, "gate") == 0 && !cd->is_gate) {
obs_data_clear(s);
obs_data_set_string(s, S_PRESETS, "gate");
expander_defaults(s);
cd->is_gate = true;
}
const uint32_t sample_rate =
audio_output_get_sample_rate(obs_get_audio());
const size_t num_channels =
audio_output_get_channels(obs_get_audio());
const float attack_time_ms =
(float)obs_data_get_int(s, S_ATTACK_TIME);
const float release_time_ms =
(float)obs_data_get_int(s, S_RELEASE_TIME);
const float output_gain_db =
(float)obs_data_get_double(s, S_OUTPUT_GAIN);
cd->ratio = (float)obs_data_get_double(s, S_RATIO);
cd->threshold = (float)obs_data_get_double(s, S_THRESHOLD);
cd->attack_gain = gain_coefficient(sample_rate,
attack_time_ms / MS_IN_S_F);
cd->release_gain = gain_coefficient(sample_rate,
release_time_ms / MS_IN_S_F);
cd->output_gain = db_to_mul(output_gain_db);
cd->num_channels = num_channels;
cd->sample_rate = sample_rate;
cd->slope = 1.0f - cd->ratio;
const char *detect_mode = obs_data_get_string(s, S_DETECTOR);
if (strcmp(detect_mode, "RMS") == 0)
cd->detector = RMS_DETECT;
if (strcmp(detect_mode, "peak") == 0)
cd->detector = PEAK_DETECT;
if (strcmp(detect_mode, "none") == 0)
cd->detector = NO_DETECT;
size_t sample_len = sample_rate * DEFAULT_AUDIO_BUF_MS / MS_IN_S;
if (cd->envelope_buf_len == 0)
resize_env_buffer(cd, sample_len);
if (cd->runaverage_len == 0)
resize_runaverage_buffer(cd, sample_len);
if (cd->maxspl_len == 0)
resize_maxspl_buffer(cd, sample_len);
if (cd->env_in_len == 0)
resize_env_in_buffer(cd, sample_len);
}
obs_output_t *obs_output_create(const char *id, const char *name,
obs_data_t *settings, obs_data_t *hotkey_data)
{
const struct obs_output_info *info = find_output(id);
struct obs_output *output;
int ret;
output = bzalloc(sizeof(struct obs_output));
pthread_mutex_init_value(&output->interleaved_mutex);
pthread_mutex_init_value(&output->delay_mutex);
if (pthread_mutex_init(&output->interleaved_mutex, NULL) != 0)
goto fail;
if (pthread_mutex_init(&output->delay_mutex, NULL) != 0)
goto fail;
if (!init_output_handlers(output, name, settings, hotkey_data))
goto fail;
if (!info) {
blog(LOG_ERROR, "Output ID '%s' not found", id);
output->info.id = bstrdup(id);
output->owns_info_id = true;
} else {
output->info = *info;
}
output->video = obs_get_video();
output->audio = obs_get_audio();
if (output->info.get_defaults)
output->info.get_defaults(output->context.settings);
ret = os_event_init(&output->reconnect_stop_event,
OS_EVENT_TYPE_MANUAL);
if (ret < 0)
goto fail;
if (info)
output->context.data = info->create(output->context.settings,
output);
if (!output->context.data)
blog(LOG_ERROR, "Failed to create output '%s'!", name);
output->reconnect_retry_sec = 2;
output->reconnect_retry_max = 20;
output->valid = true;
output->control = bzalloc(sizeof(obs_weak_output_t));
output->control->output = output;
obs_context_data_insert(&output->context,
&obs->data.outputs_mutex,
&obs->data.first_output);
blog(LOG_INFO, "output '%s' (%s) created", name, id);
return output;
fail:
obs_output_destroy(output);
return NULL;
}
static void add_audio_encoder_params(struct dstr *cmd, obs_encoder_t *aencoder)
{
obs_data_t *settings = obs_encoder_get_settings(aencoder);
int bitrate = (int)obs_data_get_int(settings, "bitrate");
audio_t *audio = obs_get_audio();
struct dstr name = {0};
obs_data_release(settings);
dstr_copy(&name, obs_encoder_get_name(aencoder));
dstr_replace(&name, "\"", "\"\"");
dstr_catf(cmd, "\"%s\" %d %d %d ",
name.array,
bitrate,
(int)obs_encoder_get_sample_rate(aencoder),
(int)audio_output_get_channels(audio));
dstr_free(&name);
}
mssapi_captions::mssapi_captions(
captions_cb callback,
const std::string &lang) try
: captions_handler(callback, AUDIO_FORMAT_16BIT, 16000)
{
HRESULT hr;
std::wstring wlang;
wlang.resize(lang.size());
for (size_t i = 0; i < lang.size(); i++)
wlang[i] = (wchar_t)lang[i];
LCID lang_id = LocaleNameToLCID(wlang.c_str(), 0);
wchar_t lang_str[32];
_snwprintf(lang_str, 31, L"language=%x", (int)lang_id);
stop = CreateEvent(nullptr, false, false, nullptr);
if (!stop.Valid())
throw "Failed to create event";
hr = SpFindBestToken(SPCAT_RECOGNIZERS, lang_str, nullptr, &token);
if (FAILED(hr))
throw HRError("SpFindBestToken failed", hr);
hr = CoCreateInstance(CLSID_SpInprocRecognizer, nullptr, CLSCTX_ALL,
__uuidof(ISpRecognizer), (void**)&recognizer);
if (FAILED(hr))
throw HRError("CoCreateInstance for recognizer failed", hr);
hr = recognizer->SetRecognizer(token);
if (FAILED(hr))
throw HRError("SetRecognizer failed", hr);
hr = recognizer->SetRecoState(SPRST_INACTIVE);
if (FAILED(hr))
throw HRError("SetRecoState(SPRST_INACTIVE) failed", hr);
hr = recognizer->CreateRecoContext(&context);
if (FAILED(hr))
throw HRError("CreateRecoContext failed", hr);
ULONGLONG interest = SPFEI(SPEI_RECOGNITION) |
SPFEI(SPEI_END_SR_STREAM);
hr = context->SetInterest(interest, interest);
if (FAILED(hr))
throw HRError("SetInterest failed", hr);
hr = context->SetNotifyWin32Event();
if (FAILED(hr))
throw HRError("SetNotifyWin32Event", hr);
notify = context->GetNotifyEventHandle();
if (notify == INVALID_HANDLE_VALUE)
throw HRError("GetNotifyEventHandle failed", E_NOINTERFACE);
size_t sample_rate = audio_output_get_sample_rate(obs_get_audio());
audio = new CaptionStream((DWORD)sample_rate, this);
audio->Release();
hr = recognizer->SetInput(audio, false);
if (FAILED(hr))
throw HRError("SetInput failed", hr);
hr = context->CreateGrammar(1, &grammar);
if (FAILED(hr))
throw HRError("CreateGrammar failed", hr);
hr = grammar->LoadDictation(nullptr, SPLO_STATIC);
if (FAILED(hr))
throw HRError("LoadDictation failed", hr);
try {
t = std::thread([this] () {main_thread();});
} catch (...) {
throw "Failed to create thread";
}
} catch (const char *err) {
blog(LOG_WARNING, "%s: %s", __FUNCTION__, err);
throw CAPTIONS_ERROR_GENERIC_FAIL;
} catch (HRError err) {
blog(LOG_WARNING, "%s: %s (%lX)", __FUNCTION__, err.str, err.hr);
throw CAPTIONS_ERROR_GENERIC_FAIL;
}
