ben.carrasco/jambonz-api-server
1
0
Fork 0
mirror of https://github.com/jambonz/jambonz-api-server.git synced 2025-12-19 05:47:46 +00:00
Code Issues Packages Projects Releases Wiki Activity

api server cannot synthesize text after upgrade latest speech-utils (#317)

Browse Source 
* api server cannot synthesize text after upgrade latest speech-utils

* wip

* add testcase for synthesize text

* fix synthesize testcase
This commit is contained in:
Hoan Luu Huu
2024-04-30 06:48:34 +07:00
committed by GitHub
parent 2436bea6ea
commit b765232d4f
2 changed files with 100 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
lib/routes/api/tts-cache.js
View File

@@ -10,6 +10,7 @@ const Account = require('../../models/account');
const sysError = require('../error');
const { getSpeechCredential, decryptCredential } = require('../../utils/speech-utils');
const PCMToMP3Encoder = require('../../record/encoder');
const { pipeline } = require('stream');
router.delete('/', async(req, res) => {
const {purgeTtsCache} = req.app.locals;
@@ -69,6 +70,8 @@ router.post('/Synthesize', async(req, res) => {
voice = arr[1];
model = arr[2];
}
} else if (cred.vendor === 'deepgram') {
model = voice;
}
const stats = {
histogram: () => {},
@@ -84,7 +87,8 @@ router.post('/Synthesize', async(req, res) => {
model,
salt,
credentials: cred,
disableTtsCache: false
disableTtsCache: false,
disableTtsStreaming: true
});
let contentType = 'audio/mpeg';
@@ -110,10 +114,17 @@ router.post('/Synthesize', async(req, res) => {
res.writeHead(200, {
'Content-Type': contentType,
});
readStream.pipe(res);
readStream.on('end', () => {
fs.unlink(filePath, (err) => {
if (err) throw err;
pipeline(readStream, res, (err) => {
if (err) {
logger.error('ttscache/Synthesize failed:', err);
if (!res.headersSent) {
res.status(500).end('Server error');
}
}
fs.unlink(filePath, (unlinkErr) => {
if (unlinkErr) throw unlinkErr;
logger.info(`${filePath} was deleted`);
});
});

84
test/speech-credentials.js
View File

@@ -170,6 +170,20 @@ test('speech credentials tests', async(t) => {
//console.log(JSON.stringify(result));
t.ok(result.statusCode === 200 && result.body.tts.status === 'ok', 'successfully tested speech credential for google tts');
t.ok(result.statusCode === 200 && result.body.stt.status === 'ok', 'successfully tested speech credential for google stt');
result = await request.post(`/Accounts/${account_sid}/TtsCache/Synthesize`, {
resolveWithFullResponse: true,
auth: authUser,
json: true,
body: {
speech_credential_sid: ms_sid,
text: "Hello How are you",
language: "en-US",
voice: "en-US-Standard-C"
}
});
t.ok(result.statusCode === 200, 'successfully google tested synthesize');
}
/* add / test a credential for microsoft */
@@ -198,6 +212,20 @@ test('speech credentials tests', async(t) => {
//console.log(JSON.stringify(result));
t.ok(result.statusCode === 200 && result.body.tts.status === 'ok', 'successfully tested speech credential for microsoft tts');
t.ok(result.statusCode === 200 && result.body.stt.status === 'ok', 'successfully tested speech credential for microsoft stt');
result = await request.post(`/Accounts/${account_sid}/TtsCache/Synthesize`, {
resolveWithFullResponse: true,
auth: authUser,
json: true,
body: {
speech_credential_sid: ms_sid,
text: "Hello How are you",
language: "en-US",
voice: "en-US-AvaMultilingualNeural"
}
});
t.ok(result.statusCode === 200, 'successfully microsoft tested synthesize');
}
/* add / test a credential for AWS */
@@ -227,6 +255,20 @@ test('speech credentials tests', async(t) => {
//console.log(JSON.stringify(result));
t.ok(result.statusCode === 200 && result.body.tts.status === 'ok', 'successfully tested speech credential for AWS tts');
t.ok(result.statusCode === 200 && result.body.stt.status === 'ok', 'successfully tested speech credential for AWS stt');
result = await request.post(`/Accounts/${account_sid}/TtsCache/Synthesize`, {
resolveWithFullResponse: true,
auth: authUser,
json: true,
body: {
speech_credential_sid: ms_sid,
text: "Hello How are you",
language: "en-US",
voice: "Joanna"
}
});
t.ok(result.statusCode === 200, 'successfully AWS tested synthesize');
}
/* add a credential for wellsaid */
@@ -253,6 +295,20 @@ test('speech credentials tests', async(t) => {
//console.log(JSON.stringify(result));
t.ok(result.statusCode === 200 && result.body.tts.status === 'ok', 'successfully tested speech credential for wellsaid');
result = await request.post(`/Accounts/${account_sid}/TtsCache/Synthesize`, {
resolveWithFullResponse: true,
auth: authUser,
json: true,
body: {
speech_credential_sid: ms_sid,
text: "Hello How are you",
language: "en-US",
voice: "3"
}
});
t.ok(result.statusCode === 200, 'successfully Wellsaid tested synthesize');
/* delete the credential */
result = await request.delete(`/Accounts/${account_sid}/SpeechCredentials/${ms_sid}`, {
auth: authUser,
@@ -285,6 +341,20 @@ test('speech credentials tests', async(t) => {
//console.log(JSON.stringify(result));
t.ok(result.statusCode === 200 && result.body.stt.status === 'ok', 'successfully tested speech credential for deepgram');
result = await request.post(`/Accounts/${account_sid}/TtsCache/Synthesize`, {
resolveWithFullResponse: true,
auth: authUser,
json: true,
body: {
speech_credential_sid: ms_sid,
text: "Hello How are you",
language: "en-US",
voice: "aura-asteria-en"
}
});
t.ok(result.statusCode === 200, 'successfully deepgram tested synthesize');
/* delete the credential */
result = await request.delete(`/Accounts/${account_sid}/SpeechCredentials/${ms_sid}`, {
auth: authUser,
@@ -404,6 +474,20 @@ test('speech credentials tests', async(t) => {
//console.log(JSON.stringify(result));
t.ok(result.statusCode === 200 && result.body.stt.status === 'ok', 'successfully tested speech credential for ibm stt');
result = await request.post(`/Accounts/${account_sid}/TtsCache/Synthesize`, {
resolveWithFullResponse: true,
auth: authUser,
json: true,
body: {
speech_credential_sid: ms_sid,
text: "Hello How are you",
language: "en-US",
voice: "en-US_MichaelExpressive"
}
});
t.ok(result.statusCode === 200, 'successfully IBM tested synthesize');
/* delete the credential */
result = await request.delete(`/Accounts/${account_sid}/SpeechCredentials/${ms_sid}`, {
auth: authUser,