How to clone your voice in 5 steps
- Record your sampleUse your phone’s voice memo app in a quiet room. Speak naturally for 15 to 30 seconds. Tips are in the next section.
- Upload it to Voice CloningOpen the tool at the top of this page and choose Upload a voice sample, or paste a direct link to the audio file. Speechma accepts clips from 5 to 30 seconds in common audio formats.
- Type what your voice should sayPaste up to 1,000 characters into Text to speak. Write numbers as you say them: “twenty twenty-six”, not “2026”.
- Pick the language and speedChoose the language of your text and a speed from 0.8x to 1.25x. Start at Normal (1x).
- Generate and downloadEnter the 5-digit security code, press Clone & Generate, listen, then download.
How to record a clean voice sample on your phone
The clone copies everything in your clip, including the room. These habits matter more than an expensive mic.
- Pick a quiet, soft room. A bedroom or a closet full of clothes beats a kitchen or bathroom. Turn off fans, AC and the TV, and close the window.
- Hold the phone 6 to 12 inches from your mouth, slightly to the side so your breath does not hit the mic. Keep the distance the same the whole time.
- Record 15 to 30 seconds. Under 5 seconds is too little; Speechma uses up to 30 seconds and trims the rest.
- One speaker only. No second voice, no kids or pets in the background, no TV.
- No music or effects. Do not use a clip from a video with a soundtrack. If you only have one, run it through the Vocal Isolator first.
- Speak the way you want the clone to sound. A calm read gives a calm clone; an energetic read gives an energetic one.
- Use varied sentences. Mix a question, a statement and a number. Avoid repeating one word.
- Trim silence and mistakes. Cut long pauses at the start and end. Keep it continuous speech.
A script to read for your sample
Not sure what to say? Read this at your normal pace. It takes about 20 seconds and covers different sounds and a question:
Hi, this is a quick test of my voice. I usually record in the morning, before the house gets busy. Do you think it will sound like me? We will find out soon. My favorite places are Chicago in the fall, the Oregon coast in summer, and my own backyard on a Sunday afternoon.
Then use the clone for what you actually need: a YouTube voiceover, a course lesson, a podcast intro, or a voicemail greeting.
Settings and supported languages
| Setting | What to choose |
|---|---|
| Voice sample | 5 to 30 seconds of clear speech; about 15 to 20 seconds usually works best |
| Text to speak | Up to 1,000 characters per generation; split longer scripts into parts |
| Language | English, Spanish, French, German, Italian, Portuguese, Russian, Chinese, Japanese or Korean |
| Speed | Slower (0.8x), Slightly slower (0.9x), Normal (1x), Slightly faster (1.1x), Faster (1.25x) |
| Output | Audio to play and download; results stay in your browser |
Your sample can be in one language and your text in another. Your voice will speak the new language, though your accent may carry over a little.
Troubleshooting: when the clone does not sound like you
- Sounds muffled or echoey: record again in a softer room, closer to the phone. Echo in the sample becomes echo in every result.
- Background hiss or hum: turn off fans and chargers near the mic, or move away from the fridge. Do not use heavy noise reduction; it makes voices sound robotic.
- Wrong accent or a different person: make sure only you speak in the clip, and try a longer, more varied sample (20 to 30 seconds).
- Too fast or too slow: change Speed, or add commas and periods to your text for natural pauses.
- Mispronounced names or numbers: spell them the way they sound: “Siobhan” as “Shi-vawn”, “2026” as “twenty twenty-six”.
- Flat or monotone: record your sample with more natural expression. The clone follows the energy you give it.
- Text cut off: stay under 1,000 characters, and generate long scripts in parts.
Your voice, your consent
This guide is for cloning your own voice. To clone anyone else, including a friend, coworker or voice actor, you need their clear, informed permission for the specific use, ideally in writing. Never clone a celebrity or public figure. See our AI Voice & Cloning Policy.
Good news for creators: YouTube lists cloning your own voice for voiceovers or dubs as a use that does not need an AI disclosure label. Your sample is processed to create the audio and is not stored on Speechma servers.
Frequently asked questions
How long should my voice sample be?
Can I record my sample on my phone?
Can my cloned voice speak another language?
Is cloning my own voice free on Speechma?
Does Speechma keep my voice?
Do I need to label videos made with my cloned voice?
Make better voiceovers, faster
- New voices and features the day they launch
- Script templates and tricks that make AI voices sound human
- Creator stories, community picks and early access
