Language Selection

Echo lets you choose which language you are speaking so the transcription model can produce the most accurate results. Depending on your model, you can select a specific language or let the model detect it automatically.

How to Select a Language

1. Open Settings and go to the Transcription tab.
2. The Language section appears below the current model display.
3. Use the dropdown to pick your language.

The language you select applies to all transcriptions until you change it.

Language Behavior by Model Type

Different model types handle language selection differently:

Multilingual Models (Whisper multilingual, cloud services)

These models support dozens of languages. You can either:

• Select a specific language from the dropdown for best accuracy.
• Choose Auto-detect to let the model determine the language from your speech.

When you select a specific language, the model focuses its recognition on that language, which tends to improve accuracy. Auto-detect is useful when you switch between languages frequently, but it may be slightly less accurate than specifying the language directly.

English-Only Models (Tiny English, Base English)

These models are optimized specifically for English. The language is fixed to English and the dropdown is not shown. You will see the label Language: English with a note that the model only supports English transcription.

Auto-Detect Models (Parakeet V2, Parakeet V3, Gemini)

Some models handle language detection entirely on their own. For these models, the language selection is disabled and you will see Language: Autodetected. The model automatically identifies the spoken language during transcription.

Supported Languages

Multilingual Whisper models and most cloud services support over 90 languages, including:

Arabic, Bengali, Chinese, Czech, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Hungarian, Indonesian, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Romanian, Russian, Spanish, Swedish, Tamil, Thai, Turkish, Ukrainian, Urdu, Vietnamese, and many more.

The full list is available in the language dropdown when a multilingual model is selected.

Apple Speech Languages

Apple Speech supports a smaller set of languages (Arabic, Chinese, Cantonese, English, French, German, Italian, Japanese, Korean, Portuguese, Spanish) but offers regional variants for each. See Apple Speech for the complete list.

Changing Language from the Menu Bar

You can also change the transcription language without opening Settings:

1. Click the Echo icon in the menu bar.
2. Look for the Language menu item.
3. Select your desired language from the list.

The selected language is indicated with a checkmark. This is the same setting as in the Transcription tab -- changing it in one place updates it everywhere.

Tips

• Setting a specific language instead of using auto-detect generally produces better results, especially for shorter recordings.
• If you switch between languages frequently, Parakeet V3 or a cloud model with auto-detect may be more convenient than manually changing the language each time.
• When you change your default model, check that the language setting still makes sense. Switching from a multilingual model to an English-only model will automatically set the language to English.

Related

• Transcription Overview -- Understanding the different transcription approaches
• Local Models -- Which local models support which languages
• Cloud Services -- Language support across cloud providers