Talk to me
Create mp3/webms from almost any type of file using MS Edge's TTS api!
Usable as a CLI tool and a library.
Works on most files, both local and remote.
Should support a lot of voices, but I've only tested with the default one.
Will probably stop working sooner rather than later! Microsoft will probably nuke this API at some point.
Based on MsEdgeTTS
Requirements
You need to have pdftotext
on your path for PDFs, antiword
for Word. See the textract
package for specific requirements.
CLI
Install
pnpm add -g read-for-me
# npm i -g read-for-me
# yarn global add read-for-me
Usage
read-for-me v0.0.1
Usage:
read-for-me [flags...] <file> [out]
Flags:
--chunk-size <number> Size of chunks to split text into. Might be worth experimenting with smaller chunks is conversion fails. (default: 20000)
--format <value> Format to use (mp3, mp3low, webm) (default: "mp3")
-h, --help Show help
--list-voices List available voices
--version Show version
--voice <value> Voice to use (default: "en-US-Jenny")
Examples
readforme somepdf.pdf
readforme whatever.txt
readforme https://apdfonline.com/somepdf.pdf --voice "en-US-Aria" hiii.mp3
API
Read the buffer yourself. If you do this, you need to pass the mime-type as well.
import { readForMe } from 'read-for-me'
import fs from 'fs/promises'
const file = await fs.readFile('somepdf.pdf')
const mp3 = await readForMe({
file,
mimeType: 'application/pdf',
format: 'mp3',
voice: 'en-US-Aria',
chunkSize: 20000,
})
You can of course also pass it a URL or location
import { readForMe } from 'read-for-me'
const mp3 = await readForMe({
file: 'https://apdfonline.com/somepdf.pdf',
format: 'mp3',
voice: 'en-US-Aria',
chunkSize: 20000,
})
Development
pnpm i