ben.carrasco/speech-utils
1
0
Fork 0
mirror of https://github.com/jambonz/speech-utils.git synced 2026-01-25 02:08:26 +00:00
Code Issues Packages Projects Releases Wiki Activity

support precache audio with tts stream enabled

Browse Source
This commit is contained in:
Quan HL
2024-08-12 18:29:01 +07:00
parent 73feadc4c4
commit b6a3fa5081
2 changed files with 33 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
lib/synth-audio.js
View File

@@ -86,7 +86,7 @@ const trimTrailingSilence = (buffer) => {
*/
async function synthAudio(client, createHash, retrieveHash, logger, stats, { account_sid,
vendor, language, voice, gender, text, engine, salt, model, credentials, deploymentId,
disableTtsCache, renderForCaching, disableTtsStreaming, options
disableTtsCache, renderForCaching, disableTtsStreaming, options, preCache = false
}) {
let audioBuffer;
let servedFromCache = false;
@@ -157,20 +157,40 @@ async function synthAudio(client, createHash, retrieveHash, logger, stats, { acc
}
}
const preCachekey = makeSynthKey({
account_sid,
vendor,
language: language || '',
voice: voice || deploymentId,
engine,
text,
preCache: true
});
const key = makeSynthKey({
account_sid,
vendor,
language: language || '',
voice: voice || deploymentId,
engine,
text
text,
preCache
});
let filePath;
filePath = makeFilePath(vendor, key, salt);
filePath = makeFilePath({vendor, key, salt, preCache});
debug(`synth key is ${key}`);
let cached;
if (!disableTtsCache) {
cached = await client.get(key);
// 1. check if precache audio
// 2. if there is no precache, check key
cached = await client.get(preCachekey);
if (cached) {
// Precache audio is available update filpath with precache file extension.
filePath = makeFilePath({vendor, key, salt, preCache: true});
} else {
cached = await client.get(key);
}
}
if (cached) {
// found in cache - extend the expiry and use it

16
lib/utils.js
View File

@@ -16,29 +16,31 @@ const debug = require('debug')('jambonz:realtimedb-helpers');
*/
//const nuanceClientMap = new Map();
function makeSynthKey({account_sid = '', vendor, language, voice, engine = '', text}) {
function makeSynthKey({
account_sid = '', vendor, language, voice, engine = '', text,
preCache = false}) {
const hash = crypto.createHash('sha1');
hash.update(`${language}:${vendor}:${voice}:${engine}:${text}`);
const hexHashKey = hash.digest('hex');
const accountKey = account_sid ? `:${account_sid}` : '';
const namespace = vendor.startsWith('custom') ? vendor : getFileExtension(vendor);
const namespace = vendor.startsWith('custom') ? vendor : getFileExtension({vendor, preCache});
const key = `tts${accountKey}:${namespace}:${hexHashKey}`;
return key;
}
function makeFilePath(vendor, key, salt = '') {
const extension = getFileExtension(vendor);
function makeFilePath({vendor, key, salt = '', preCache = false}) {
const extension = getFileExtension({vendor, preCache});
return `${TMP_FOLDER}/${key.replace('tts:', `tts-${salt}`)}.${extension}`;
}
function getFileExtension(vendor) {
function getFileExtension({vendor, preCache = false}) {
const mp3Extension = 'mp3';
const r8Extension = 'r8';
switch (vendor) {
case 'azure':
case 'microsoft':
if (!JAMBONES_DISABLE_TTS_STREAMING || JAMBONES_TTS_TRIM_SILENCE) {
if (!preCache && !JAMBONES_DISABLE_TTS_STREAMING || JAMBONES_TTS_TRIM_SILENCE) {
return r8Extension;
} else {
return mp3Extension;
@@ -46,7 +48,7 @@ function getFileExtension(vendor) {
case 'deepgram':
case 'elevenlabs':
case 'rimlabs':
if (!JAMBONES_DISABLE_TTS_STREAMING) {
if (!preCache && !JAMBONES_DISABLE_TTS_STREAMING) {
return r8Extension;
} else {
return mp3Extension;