ben.carrasco/speech-utils
1
0
Fork 0
mirror of https://github.com/jambonz/speech-utils.git synced 2025-12-19 03:37:49 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #19 from jambonz/fix/nvidia

Browse Source 
fixes for riva tts
This commit is contained in:
Dave Horton
2023-05-10 07:38:58 -04:00
committed by GitHub
parent d606141f57 b6058ca242
commit 4349cd7e40
2 changed files with 18 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
lib/synth-audio.js
View File

@@ -81,7 +81,7 @@ async function synthAudio(client, logger, stats, { account_sid,
else if ('nvidia' === vendor) {
assert.ok(voice, 'synthAudio requires voice when nvidia is used');
assert.ok(language, 'synthAudio requires language when nvidia is used');
assert.ok(credentials.riva_uri, 'synthAudio requires riva_uri in credentials when nuance is used');
assert.ok(credentials.riva_server_uri, 'synthAudio requires riva_server_uri in credentials when nvidia is used');
}
else if ('ibm' === vendor) {
assert.ok(voice, 'synthAudio requires voice when ibm is used');
@@ -425,20 +425,25 @@ const synthNuance = async(client, logger, {credentials, stats, voice, model, tex
};
const synthNvidia = async(client, logger, {credentials, stats, language, voice, model, text}) => {
const {riva_uri} = credentials;
const rivaClient = await createRivaClient(riva_uri);
const request = new SynthesizeSpeechRequest();
request.setVoiceName(voice);
request.setLanguageCode(language);
request.setSampleRateHz(8000);
request.setEncoding(AudioEncoding.LINEAR_PCM);
request.setText(text);
const {riva_server_uri} = credentials;
let rivaClient, request;
try {
rivaClient = await createRivaClient(riva_server_uri);
request = new SynthesizeSpeechRequest();
request.setVoiceName(voice);
request.setLanguageCode(language);
request.setSampleRateHz(8000);
request.setEncoding(AudioEncoding.LINEAR_PCM);
request.setText(text);
} catch (err) {
logger.info({err}, 'error creating riva client');
return Promise.reject(err);
}
return new Promise((resolve, reject) => {
rivaClient.synthesize(request, (err, response) => {
if (err) {
console.error(err);
logger.info({err, voice, language}, 'error synthesizing speech using Nvidia');
return reject(err);
}
resolve(Buffer.from(response.getAudio()));

4
test/synth.js
View File

@@ -300,7 +300,7 @@ test('Nvidia speech synth tests', async(t) => {
let opts = await synthAudio(stats, {
vendor: 'nvidia',
credentials: {
riva_uri: process.env.RIVA_URI,
riva_server_uri: process.env.RIVA_URI,
},
language: 'en-US',
voice: 'English-US.Female-1',
@@ -311,7 +311,7 @@ test('Nvidia speech synth tests', async(t) => {
opts = await synthAudio(stats, {
vendor: 'nvidia',
credentials: {
riva_uri: process.env.RIVA_URI,
riva_server_uri: process.env.RIVA_URI,
},
language: 'en-US',
voice: 'English-US.Female-1',