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

Set language/rate in prosody XML

Browse source
This commit is contained in:
Jackson Harper 2022-08-29 17:20:43 +08:00
parent e298a49169
commit 4f6d8cdfce
3 changed files with 47 additions and 20 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

23
packages/text-to-speech/src/htmlToSsml.ts
View file

@ -131,16 +131,20 @@ export type SSMLItem = {
textItems: string[]
}
export type VoiceOptions = {
primary: string
secondary: string
export type SSMLOptions = {
primaryVoice: string
secondaryVoice: string
rate: string
language: string
}
const startSsml = (element: Element, voices: VoiceOptions): string => {
const startSsml = (element: Element, options: SSMLOptions): string => {
const voice =
element.nodeName === 'BLOCKQUOTE' ? voices.secondary : voices.primary
element.nodeName === 'BLOCKQUOTE'
? options.secondaryVoice
: options.primaryVoice
return `
<speak xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="http://www.w3.org/2001/mstts" xmlns:emo="http://www.w3.org/2009/10/emotionml" version="1.0" xml:lang="en-US"><voice name="${voice}"><prosody rate="0%" pitch="0%">
<speak xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="http://www.w3.org/2001/mstts" xmlns:emo="http://www.w3.org/2009/10/emotionml" version="1.0" xml:lang="${options.language}"><voice name="${voice}"><prosody rate="${options.rate}" pitch="default">
`
}
@ -152,10 +156,7 @@ export const ssmlItemText = (item: SSMLItem): string => {
return [item.open, ...item.textItems, item.close].join('')
}
export const htmlToSsml = (
html: string,
voices: { primary: string; secondary: string }
): SSMLItem[] => {
export const htmlToSsml = (html: string, options: SSMLOptions): SSMLItem[] => {
const dom = parseHTML(html)
const body = dom.document.querySelector('#readability-page-1')
if (!body) {
@ -174,7 +175,7 @@ export const htmlToSsml = (
i = emitElement(textItems, node, true)
items.push({
open: startSsml(node, voices),
open: startSsml(node, options),
close: endSsml(),
textItems: textItems,
})

6
packages/text-to-speech/src/index.ts
View file

@ -217,8 +217,10 @@ const synthesizeTextToSpeech = async (
}
} else {
const ssmlItems = htmlToSsml(input.text, {
primary: speechConfig.speechSynthesisVoiceName,
secondary: 'en-US-GuyNeural',
primaryVoice: speechConfig.speechSynthesisVoiceName,
secondaryVoice: 'en-US-GuyNeural',
language: speechConfig.speechSynthesisLanguage,
rate: '1',
})
for (const ssmlItem of Array.from(ssmlItems)) {

38
packages/text-to-speech/test/htmlToSsml.test.ts
View file

@ -1,9 +1,17 @@
import 'mocha'
import { expect } from 'chai'
import fs from 'fs'
import { glob } from 'glob'
import { htmlToSsml } from '../src/htmlToSsml'
describe('htmlToSsml', () => {
const TEST_VOCIES = { primary: 'test-primary', secondary: 'test-secondary' }
const TEST_OPTIONS = {
primaryVoice: 'test-primary',
secondaryVoice: 'test-secondary',
language: 'en-US',
rate: '1'
}
describe('a simple html file', () => {
it('should convert Html to SSML', async () => {
@ -11,7 +19,7 @@ describe('htmlToSsml', () => {
<div class="page" id="readability-page-1">
<p data-omnivore-anchor-idx="1">this is some text</p>
</div>
`, TEST_VOCIES
`, TEST_OPTIONS
)
const text = ssml[0].textItems.join('').trim()
expect(text).to.equal(
@ -29,7 +37,7 @@ this is in the first paragraph
this is also in the first paragraph
</p>
</div>
`, TEST_VOCIES
`, TEST_OPTIONS
)
const text = ssml[0].textItems.join('').trim()
expect(text).to.equal(
@ -45,7 +53,7 @@ this is also in the first paragraph
<blockquote>second</blockquote>
<p>third</p>
</div>
`, TEST_VOCIES
`, TEST_OPTIONS
)
const first = ssml[0].textItems.join('').trim()
const second = ssml[1].textItems.join('').trim()
@ -62,14 +70,30 @@ this is also in the first paragraph
)
expect(ssml[0].open.trim()).to.equal(
`<speak xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="http://www.w3.org/2001/mstts" xmlns:emo="http://www.w3.org/2009/10/emotionml" version="1.0" xml:lang="en-US"><voice name="test-primary"><prosody rate="0%" pitch="0%">`
`<speak xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="http://www.w3.org/2001/mstts" xmlns:emo="http://www.w3.org/2009/10/emotionml" version="1.0" xml:lang="en-US"><voice name="test-primary"><prosody rate="1" pitch="default">`
)
expect(ssml[1].open.trim()).to.equal(
`<speak xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="http://www.w3.org/2001/mstts" xmlns:emo="http://www.w3.org/2009/10/emotionml" version="1.0" xml:lang="en-US"><voice name="test-secondary"><prosody rate="0%" pitch="0%">`
`<speak xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="http://www.w3.org/2001/mstts" xmlns:emo="http://www.w3.org/2009/10/emotionml" version="1.0" xml:lang="en-US"><voice name="test-secondary"><prosody rate="1" pitch="default">`
)
expect(ssml[2].open.trim()).to.equal(
`<speak xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="http://www.w3.org/2001/mstts" xmlns:emo="http://www.w3.org/2009/10/emotionml" version="1.0" xml:lang="en-US"><voice name="test-primary"><prosody rate="0%" pitch="0%">`
`<speak xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="http://www.w3.org/2001/mstts" xmlns:emo="http://www.w3.org/2009/10/emotionml" version="1.0" xml:lang="en-US"><voice name="test-primary"><prosody rate="1" pitch="default">`
)
})
})
// For local testing:
// describe('readability test files', () => {
// it('should convert Html to SSML without throwing', async () => {
// const g = new glob.GlobSync('../readabilityjs/test/test-pages/*')
// console.log('glob: ', glob)
// for (const f of g.found) {
// const readablePath = `${f}/expected.html`
// if (!fs.existsSync(readablePath)) {
// continue
// }
// const html = fs.readFileSync(readablePath, { encoding: 'utf-8' })
// const ssmlItems = htmlToSsml(html, TEST_OPTIONS)
// console.log('SSML ITEMS', ssmlItems)
// }
// })
// })
})