Last month, recording my third course video, my voice gave out again
I make paid courses on the side, recording two talking-head segments a week. That night, after recording the third one, my voice was shot. Staring at the camera, I thought: I can't keep grinding like this.
You probably know this too — don't want to show your face but need a voice, want to make courses but don't have time to set up a real studio. AI voice tools have changed a lot in the past two years. The 2025 versions genuinely sound 'un-detectable as machine.'
Three tools — I used each for two months
Bottom line up front: for course recording, ElevenLabs is the best deal (starts at $5/month, about ¥36 RMB); for daily short-form video with lower demands, JianYing's built-in AI voice is enough, free; for commercial voiceover gigs with high detail requirements, go with ElevenLabs' voice cloning tier or Microsoft Azure.
My friend Zhou runs finance short-form video in Hangzhou, team of two — him and an editor. Since March 2025 he's been using ElevenLabs' 'clone your own voice' feature — first he records a 30-minute sample to feed the system, then for every video he just types text and only needs to double-check occasional mispronounced words himself. He says the saved voice lets him post two extra videos a week.
Another friend, Sister Chen, makes children's audiobooks in Chengdu. She tried JianYing, Moyin Workshop, then settled on ElevenLabs' standard paid plan — around $22/month. She says what matters most isn't 'does it sound human' but 'can I manually fix mispronounced names' — that's something free tools can't handle.
Want to try today? Here's the replication cost
Money: JianYing is free; ElevenLabs starts at $5/month; Microsoft Azure is billed per character, with a free tier for new users for the first 12 months.
Time: First session, 2-3 hours to get the hang of it; cloning your own voice needs another 20-30 minutes of recording samples.
Tech barrier: No coding needed — all mouse-clicks, type text, download audio file. If you've used JianYing, you can use these.
First step: Go to ElevenLabs' website, log in with Google account, and you'll see 'Text to Speech' right on the homepage. Type some Chinese to test. You can even listen to demos without signing up.
Here's where I got stuck too: in ElevenLabs' Chinese voices, dialects and er-hua (rhotic) sounds occasionally get mispronounced, especially with people's names. Don't jump straight into course production on your first try — test with an old script first.
Which category are you in?
If you're just starting and haven't started making videos yet: Don't rush. It's fine to skip this now — record two videos on your phone first to see what feedback you get. When you actually feel 'my voice isn't enough' or 'I'm awkward on camera,' then consider AI voice. Not everyone needs this tool.
If you already have 1-2 clients and are starting content delivery: Start with JianYing's AI voice, free trial for a week. If after a week you feel 'I could be a bit more professional,' upgrade to ElevenLabs' $5 tier.
If you're already scaling — taking commercial voiceover or course orders: Go straight to ElevenLabs' $22 tier or voice cloning, lock down the 'record-review-export' workflow. Each saved hour adds up visibly.
One last honest word: AI voice can solve 'voice running out' and 'not wanting to show face,' but it can't solve 'is the content itself worth listening to.' The tool is an amplifier, not an engine.