prehistoric-systems/omnivore
1
0
Fork 0
mirror of https://github.com/omnivore-app/omnivore.git synced 2026-03-11 08:54:26 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Return an array of SSMLs instead of ssmlItems

Browse source
This commit is contained in:
Hongbo Wu 2022-09-02 13:51:20 +08:00
parent 71144d3c84
commit 9826262ef4
5 changed files with 44 additions and 41 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

8
packages/api/src/textToSpeech.d.ts vendored
View file

@ -1,5 +1,5 @@
declare module '@omnivore/text-to-speech-handler' {
function htmlToSsml(html: string, options: SSMLOptions): SSMLItem[]
function htmlToSsml(html: string, options: SSMLOptions): string[]
interface SSMLOptions {
primaryVoice: string
@ -8,11 +8,5 @@ declare module '@omnivore/text-to-speech-handler' {
language: string
}
interface SSMLItem {
open: string
close: string
textItems: string[]
}
export { htmlToSsml }
}

9
packages/text-to-speech/src/htmlToSsml.ts
View file

@ -179,7 +179,10 @@ export const ssmlItemText = (item: SSMLItem): string => {
return [item.open, ...item.textItems, item.close].join('')
}
export const htmlToSsml = (html: string, options: SSMLOptions): SSMLItem[] => {
export const htmlToSsmlItems = (
html: string,
options: SSMLOptions
): SSMLItem[] => {
console.log('creating ssml with options', options)
const dom = parseHTML(html)
@ -218,3 +221,7 @@ export const htmlToSsml = (html: string, options: SSMLOptions): SSMLItem[] => {
return items
}
export const htmlToSsml = (html: string, options: SSMLOptions): string[] => {
return htmlToSsmlItems(html, options).map(ssmlItemText)
}

7
packages/text-to-speech/src/index.ts
View file

@ -10,7 +10,7 @@ import * as dotenv from 'dotenv' // see https://github.com/motdotla/dotenv#how-d
import { synthesizeTextToSpeech, TextToSpeechInput } from './textToSpeech'
import { File, Storage } from '@google-cloud/storage'
import { PassThrough } from 'stream'
import { htmlToSsml, SSMLItem } from './htmlToSsml'
import { htmlToSsml } from './htmlToSsml'
dotenv.config()
Sentry.GCPFunction.init({
@ -141,7 +141,10 @@ export const textToSpeechStreamingHandler = Sentry.GCPFunction.wrapHttpFunction(
try {
// eslint-disable-next-line @typescript-eslint/no-unsafe-member-access
const ssmlItems = req.body.ssmlItems as SSMLItem[]
const ssmlItems = req.body.ssmlItems as string[]
if (!ssmlItems || ssmlItems.length === 0) {
return res.status(200).send({ errorCode: 'INVALID_DATA' })
}
const audioStream = new PassThrough()
const input: TextToSpeechInput = {
text: '',

39
packages/text-to-speech/src/textToSpeech.ts
View file

@ -7,7 +7,7 @@ import {
SpeechSynthesisResult,
SpeechSynthesizer,
} from 'microsoft-cognitiveservices-speech-sdk'
import { htmlToSsml, SSMLItem, ssmlItemText } from './htmlToSsml'
import { htmlToSsmlItems, ssmlItemText } from './htmlToSsml'
export interface TextToSpeechInput {
id?: string
@ -20,7 +20,7 @@ export interface TextToSpeechInput {
complimentaryVoice?: string
bucket?: string
audioStream: NodeJS.ReadWriteStream
ssmlItems?: SSMLItem[]
ssmlItems?: string[]
}
export interface TextToSpeechOutput {
@ -41,6 +41,7 @@ export const synthesizeTextToSpeech = async (
if (!process.env.AZURE_SPEECH_KEY || !process.env.AZURE_SPEECH_REGION) {
throw new Error('Azure Speech Key or Region not set')
}
const textType = input.textType || 'html'
const writeStream = input.audioStream
const speechConfig = SpeechConfig.fromSubscription(
process.env.AZURE_SPEECH_KEY,
@ -131,24 +132,22 @@ export const synthesizeTextToSpeech = async (
}
try {
const ssmlItems =
input.textType === 'ssml'
? input.ssmlItems
: htmlToSsml(input.text, {
primaryVoice: input.voice || 'en-US-JennyNeural',
secondaryVoice: input.complimentaryVoice || 'en-US-GuyNeural',
language: input.languageCode || 'en-US',
rate: '1',
})
if (!ssmlItems || ssmlItems.length === 0) {
throw new Error('No SSML items found')
}
for (const ssmlItem of Array.from(ssmlItems)) {
const ssml = ssmlItemText(ssmlItem)
console.debug('start synthesizing', ssml)
const result = await speakSsmlAsyncPromise(ssml)
timeOffset = timeOffset + result.audioDuration
if (textType === 'html') {
const ssmlItems = htmlToSsmlItems(input.text, {
primaryVoice: input.voice || 'en-US-JennyNeural',
secondaryVoice: input.complimentaryVoice || 'en-US-GuyNeural',
language: input.languageCode || 'en-US',
rate: '1',
})
for (const ssmlItem of ssmlItems) {
const ssml = ssmlItemText(ssmlItem)
const result = await speakSsmlAsyncPromise(ssml)
timeOffset = timeOffset + result.audioDuration
}
} else {
for (const ssmlItem of input.ssmlItems || []) {
await speakSsmlAsyncPromise(ssmlItem)
}
}
} catch (error) {
console.error('synthesis error', error)

22
packages/text-to-speech/test/htmlToSsml.test.ts
View file

@ -1,8 +1,8 @@
import 'mocha'
import { expect } from 'chai'
import { htmlToSsml } from '../src/htmlToSsml'
import { htmlToSsmlItems } from '../src/htmlToSsml'
describe('htmlToSsml', () => {
describe('htmlToSsmlItems', () => {
const TEST_OPTIONS = {
primaryVoice: 'test-primary',
secondaryVoice: 'test-secondary',
@ -12,7 +12,7 @@ describe('htmlToSsml', () => {
describe('a simple html file', () => {
xit('should convert Html to SSML', () => {
const ssml = htmlToSsml(
const ssml = htmlToSsmlItems(
`
<div id="readability-content">
<div id="readability-page-1">
@ -28,7 +28,7 @@ describe('htmlToSsml', () => {
})
describe('a file with nested elements', () => {
xit('should collapse spans into the parent paragraph', () => {
const ssml = htmlToSsml(
const ssml = htmlToSsmlItems(
`
<div id="readability-content">
<div class="page" id="readability-page-1">
@ -57,7 +57,7 @@ describe('htmlToSsml', () => {
)
})
xit('should extract child paragraphs to the top level', () => {
const ssml = htmlToSsml(
const ssml = htmlToSsmlItems(
`
<div id="readability-content">
<div>
@ -77,7 +77,7 @@ describe('htmlToSsml', () => {
)
})
xit('should hoist paragraphs in spans to the top level', () => {
const ssml = htmlToSsml(
const ssml = htmlToSsmlItems(
`
<div id="readability-content">
<div>
@ -95,7 +95,7 @@ describe('htmlToSsml', () => {
expect(text).to.equal(`TBD`.trim())
})
xit('should hoist lists to the top level', () => {
const ssml = htmlToSsml(
const ssml = htmlToSsmlItems(
`
<div id="readability-content">
<div>
@ -113,7 +113,7 @@ describe('htmlToSsml', () => {
expect(text).to.equal(`TBD`.trim())
})
xit('should hoist headers to the top level', () => {
const ssml = htmlToSsml(
const ssml = htmlToSsmlItems(
`
<div id="readability-content">
<div>
@ -131,7 +131,7 @@ describe('htmlToSsml', () => {
expect(text).to.equal(`TBD`.trim())
})
xit('should hoist blockquotes to the top level', () => {
const ssml = htmlToSsml(
const ssml = htmlToSsmlItems(
`
<div id="readability-content">
<div>
@ -151,7 +151,7 @@ describe('htmlToSsml', () => {
})
describe('a file with blockquotes', () => {
xit('should convert Html to SSML with complimentary voices', () => {
const ssml = htmlToSsml(
const ssml = htmlToSsmlItems(
`
<div id="readability-content">
<div class="page" id="readability-page-1">
@ -193,7 +193,7 @@ describe('htmlToSsml', () => {
// continue
// }
// const html = fs.readFileSync(readablePath, { encoding: 'utf-8' })
// const ssmlItems = htmlToSsml(html, TEST_OPTIONS)
// const ssmlItems = htmlToSsmlItems(html, TEST_OPTIONS)
// console.log('SSML ITEMS', ssmlItems)
// }
// })