misc changes

pnfo · Jul 25, 2023 · a86de0c · a86de0c
1 parent 49a55dc
commit a86de0c
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 7 deletions.
diff --git a/dev-prompts/combine-lists.js b/dev-prompts/combine-lists.js
@@ -4,7 +4,7 @@ import fs from 'fs'
 import {loadPrompts} from '../common_functions.js'
 import lodash from 'lodash'
 
-const rootFolder = 'sinhala-prompts', 
+const rootFolder = 'dev-prompts', 
     textLengthToTimeRatio = 0.1035, // median from 11 test recordings from Ven mettananada 
     maxPromptLength = 13 / textLengthToTimeRatio
 

diff --git a/docs/index.html b/docs/index.html
@@ -24,7 +24,7 @@ <h1>Sinhala Text to Speech Demo</h1>
             The training took about 24 hours.</p>
         <p>The trained model is included in the releases section of this repository <a href="https://github.com/pathnirvana/coqui-tts/releases">here</a></p>
         <p>Example texts for the following voice synthesis come from the <a href="https://tipitaka.lk/">Buddha Jayanthi Tipiṭaka</a> and <a href="https://pitaka.lk/books/app-index.html">Buddhist Books App</a></p>
-        <p>TODO: Currently even thought there are two speakers in the dataset only the male voice was used for the training since when trained with both speakers' voices the synthesis quality was not good. 
+        <p>TODO: Currently even though there are two speakers in the dataset only the male voice was used for the training since when trained with both speakers' voices the synthesis quality was not good. 
       </div>
       <div class="audio-group">
         <h2>Buddhist Text</h2>

diff --git a/prompt-creator-sinhala.js → prompt-creator.js b/prompt-creator-sinhala.js → prompt-creator.js
@@ -80,8 +80,8 @@ const minSplitPromptLength = 10, maxTimeNeeded = 3600 * 20,
 console.log(`max prompt length ${maxPromptLength}, dataset name ${datasetName}`)
 const selected = {}, selectedFiles = {}, selectedTypes = {}
 let countSelected = 0, timeSelected = 0, filesUsed = 0, promptsConsidered = 0, entriesConsidered = 0
-const allSyls = JSON.parse(fs.readFileSync(`sinhala-prompts/all-syls.json`, 'utf-8'))
-const syls = JSON.parse(fs.readFileSync(`sinhala-prompts/syls-syl-1.json`, 'utf-8'))  //{select: {}, need: {}} //
+const allSyls = JSON.parse(fs.readFileSync(`dev-prompts/all-syls.json`, 'utf-8'))
+const syls = JSON.parse(fs.readFileSync(`dev-prompts/syls-syl-1.json`, 'utf-8'))  //{select: {}, need: {}} //
 
 const files = fs.readdirSync(textInputFolder).filter(f => f.endsWith('json') && !f.startsWith('atta') && !f.startsWith('anya'))
 lodash.shuffle(files).forEach(file => {
@@ -94,12 +94,12 @@ lodash.shuffle(files).forEach(file => {
 })
 
 syls.need = Object.fromEntries(Object.entries(allSyls).filter(([s, c]) => !syls.select[s] && c >= 2))
-fs.writeFileSync(`sinhala-prompts/syls-${datasetName}.json`, jsb(syls, null, '\t', 100), 'utf-8')
-fs.writeFileSync(`sinhala-prompts/prompts-${datasetName}.txt`, Object.entries(selected)
+fs.writeFileSync(`dev-prompts/syls-${datasetName}.json`, jsb(syls, null, '\t', 100), 'utf-8')
+fs.writeFileSync(`dev-prompts/prompts-${datasetName}.txt`, Object.entries(selected)
     .sort((a, b) => a[1].type.localeCompare(b[1].type))
     .map(([text, {type, file, length}], i) => `${i + 1}\t${type}\t${file.slice(0, -5)}\t${length}\n${text.replace(/ x /g, '\n')}`).join('\n\n'), 'utf-8')
 
-// fs.writeFileSync(`sinhala-prompts/prompts-common.txt`, Object.entries(allPrompts)
+// fs.writeFileSync(`dev-prompts/prompts-common.txt`, Object.entries(allPrompts)
 //     .sort((a, b) => sortLength(b) - sortLength(a)).slice(0, 1000).sort((a, b) => a[0].localeCompare(b[0]))
 //     .map(([text, count], i) => [i, count, sortLength([text, count]), text].join('\t')).join('\n'), 'utf-8')