Translate a WhatsApp voice message and reply in their language

Example: you are renting a flat in Lisbon, and the owner sends you a 40-second voice message in Portuguese. You want to understand it and answer with a voice message in Portuguese, without typing. Everything below happens in the VocaLingo app on the same phone as WhatsApp.

Which tool to use and why

Translate Audio fits this task: it accepts the voice message straight from WhatsApp, detects the language itself and starts the translation without a button. Its Record a reply button sets up the reverse direction for you, from your language back to Portuguese.

Step 1. Understand the message

1. In WhatsApp, touch and hold the voice message. On iPhone tap Forward and then the share icon in the bottom right corner; on Android tap the share icon at the top.

2. Choose Translate Audio in VocaLingo.

3. VocaLingo recognizes the Portuguese and translates it into your app language. Read it on the Translation tab, and check the original on Source text if a word looks odd.

If VocaLingo isn't in the share menu, or you use other messengers, see Send a voice message from WhatsApp or Telegram to VocaLingo.

Step 2. Record the reply

1. Under the translation, in the What next? block, tap Record a reply. Translate Audio opens with Portuguese as the target language, and recording starts right away.

2. Say your answer in your own language and tap Done.

3. Check the Portuguese text on the Translation tab. If something is wrong, tap Edit on the Source text tab and fix your words: the translation updates by itself.

4. Tap the play button. Portuguese is one of the 10 languages where the app clones your voice right away, so the answer sounds like you.

Step 3. Send the answer back

1. Tap the share button in the player.

2. Choose WhatsApp and the chat with the owner. The answer arrives as an audio file.

If you'd rather send text, copy the translation from the Translation tab and paste it into the chat.

How much it costs and how long it takes

For a short exchange like this you spend a few tokens. Recognition costs about 0.7 tokens per minute, translation fractions of a token, and the reply voiced with a Qwen 3 clone costs about 20 tokens per 1,000 characters, at least 5 tokens. A reply of 20 seconds is usually about 300 characters.

The wait depends on the length of the recording. Settings show about 3 seconds for recognition and about 20 seconds for a Qwen 3 clone, plus the time to upload the file.

If the language is not one of the 10

For languages outside English, Russian, Chinese, German, French, Spanish, Italian, Japanese, Korean and Portuguese, the app asks Clone voice? before the voiceover and shows the price. MiniMax cloning costs 180 tokens the first time. To pay it once, tap Save voice after the first cloned reply. If the voice sounds wrong, see Why the translated voice sounds robotic.

Frequently asked questions