Feat/ambient sounds (#678)

* initial support for coaching mode in conference

* wip

* wip

* add support for answer verb

* wip

* wip

* wip

* wip

* wip

* updates to rename option to dub

* wip

* wip

* wip

* update verb-specs

* wip

* wip

* wip

* wip

* wip

* wip

* wip

* wip

* add option to boost audio signal in main channel

* wip

* wip

* wip

* wip

* wip

* wip

* for now, bypass use of streaming apis when generating tts audio for dub tracks

* add nested dub to dial

* wip

* add support for filler noise

* kill filler noise when gather killed

* wip

* wip

* while using sayOnTrack, we have to enclose the say command in double quotes

* disableTtsStreaming = false

* allow transcribe of b leg only on dial verb

* dub.say can either be text or object like say verb with text and synthesizer

* remove loop for sayOnTrack

* update speech-utils

* fixes for testing transcribe verb and support for dub and boostAudioSignal in lcc commands

* add dial.boostAudioSignal

* fix bug where session-level recognizer settings incorrectly overwrite verb-level settings

* update verb specs

* update dial to support array of dub verbs

* fix bug setting gain

* lint

* wip

* update speech-utils

* use new endpoint methods for mod_dub

---------

Co-authored-by: Dave Horton <daveh@beachdognet.com>
This commit is contained in:
Hoan Luu Huu
2024-03-24 03:23:57 +07:00
committed by GitHub
parent ec58232b61
commit 5b1d8a8ff3
18 changed files with 915 additions and 278 deletions

View File

@@ -56,24 +56,20 @@ class SttTask extends Task {
super.exec(cs);
this.ep = ep;
this.ep2 = ep2;
// copy all value from config verb to this object.
// use session preferences if we don't have specific verb-level settings.
if (cs.recognizer) {
for (const k in cs.recognizer) {
if (Array.isArray(this.data.recognizer[k]) ||
Array.isArray(cs.recognizer[k])) {
this.data.recognizer[k] = [
...this.data.recognizer[k],
...cs.recognizer[k]
];
} else if (typeof this.data.recognizer[k] === 'object' ||
typeof cs.recognizer[k] === 'object'
) {
this.data.recognizer[k] = {
...this.data.recognizer[k],
...cs.recognizer[k]
};
const newValue = this.data.recognizer && this.data.recognizer[k] !== undefined ?
this.data.recognizer[k] :
cs.recognizer[k];
if (Array.isArray(newValue)) {
this.data.recognizer[k] = [...(this.data.recognizer[k] || []), ...cs.recognizer[k]];
} else if (typeof newValue === 'object' && newValue !== null) {
this.data.recognizer[k] = { ...(this.data.recognizer[k] || {}), ...cs.recognizer[k] };
} else {
this.data.recognizer[k] = cs.recognizer[k] || this.data.recognizer[k];
this.data.recognizer[k] = newValue;
}
}
}