How to Use Altered to Turn Your Voice Into Studio-Grade Audio
Altered Studio is a browser tool for changing voices: record or upload audio, pick a target voice or preset, and export. It handles voice conversion, dubbing into another language, and per-line voice edits for podcasts and games.
The interesting part is granularity. You can keep the timing and pitch of the original performance while swapping the timbre, which is what makes dubbing sound like acting instead of text-to-speech.
What can Altered actually change about a voice?
Three things, and they are separable:
- Timbre — whose voice it sounds like. This is the main feature.
- Pitch and formants — how high or low, adjusted independently.
- Timing — kept from the source unless you force a different pacing.
Separating these is the difference between a convincing voice change and the robot effect you get from a simple pitch shifter.
Presets cover the usual cases: male to female, female to male, child, elderly, and a set of character voices. Custom voice models can be trained from a few minutes of clean audio.
How do you process a file with Altered step by step?
- Sign in at altered.ai and open the studio in the browser. Chrome or Edge only; Safari has recording issues.
- Upload a WAV, MP3 or MP4, or record straight into the timeline.
- Select the region you want to change. You can process one sentence and leave the rest alone.
- Open the Voice Changer panel and choose a preset or upload a custom voice.
- Preview. Adjust pitch and formant sliders before committing.
- Export as WAV 48 kHz for editing, or MP3 for distribution.
A 3-minute voice clip processes in roughly 20-40 seconds on the built-in engine. Longer files queue.
How much does Altered cost in 2026?
There is a limited free tier for testing, with usage metered in minutes and a watermark or quality cap. Paid plans start in the low tens of dollars per month and scale by processing minutes.
Custom voice training sits on higher tiers. If your workflow is one podcast episode a week at 45 minutes, budget for the tier above entry — the entry plan will run out mid-month.
Commercial licensing matters here. Check that your plan covers monetised output before you publish. Game studios and ad agencies need written terms; the standard consumer plan is not written for that.
What are the real limits?
Background noise is the big one. Altered converts the voice; it also converts the noise that sits in the same frequency band. Record dry, or the result sounds like a voice change applied to a recording of a room.
Emotional range is the second. Shouting and crying both break conversion. The model handles conversational and dramatic delivery well and clips on extremes.
Third: consent. Do not use someone’s voice without permission. Altered’s terms require it, and in the EU the AI Act adds disclosure duties for synthetic voice in most contexts.
Who is this for, and who should skip it?
Fits — indie podcasters who want one host voice across characters, game devs prototyping NPC lines before hiring actors, e-learning teams localising a course into four languages, video editors fixing a line the actor flubbed.
Does not fit — live streaming real-time voice change (use a desktop tool with low latency), music vocals with wide vibrato, and anyone who needs an offline or air-gapped workflow. This is a web app.
If you are generating voice from text rather than converting existing audio, a different category works better. And if you need to edit the video track around the audio, Blizzy covers the video side.
Quick Summary
- Altered Studio converts voice timbre while keeping the source performance’s timing and delivery.
- Runs in Chrome or Edge; WAV, MP3 and MP4 input, 48 kHz WAV export.
- Free tier is for testing; paid plans meter processing minutes, so estimate your monthly audio volume first.
- Clean, dry recordings convert well. Noisy rooms carry over the noise.
- Shouting and crying break conversion; conversational and dramatic speech work.
