Transcribing and translating audio with Lara (audio-to-text)

Lara supports audio-to-text translation, allowing you to upload an audio file and receive both a transcription of the original speech and its translation as text output.

This feature is useful when you need the content of an audio file in written form - for example, to generate subtitles, produce translated transcripts, or feed spoken content into text-based workflows.

How audio-to-text translation works

When you submit an audio file for audio-to-text translation:

1. Lara transcribes the spoken content from the source audio
2. The transcribed text is translated into the selected target language
3. The result is returned as structured text, containing both the original transcription and its translation, organized by paragraph
   
   

How audio-to-text differs from audio-to-audio

Lara supports two distinct audio translation workflows:

• Audio-to-audio: the output is a translated audio file. The voice is synthesized in the target language.
• Audio-to-text: the output is text. You receive the transcription of the original speech and its translation.

Use audio-to-text when you need a written output, and audio-to-audio when you need a translated voice file.

Supported languages

The source audio can be in any of the 68 languages supported for audio input. For the full list, refer to:
https://developers.laratranslate.com/docs/translate-audio

The translated text output supports over 200 languages.

Supported audio formats

Audio-to-text translation accepts the following input formats:

• .wav
• .mp3
• .opus
• .ogg
• .webm

File requirements

• Maximum file size: 200 MB
• Maximum duration: 2 hours
  
  

Customization options

Audio-to-text translation supports the same adaptation options available for other Lara translation services:

• Translation memories: apply your existing TMs to ensure consistency with previously translated content
• Glossaries: enforce preferred terminology in the text output
• Translation style: choose between Faithful, Fluid, or Creative
• Incognito mode: if enabled, source content and its translation will not be stored

Usage and pricing

Audio-to-text translation usage is calculated based on audio duration.

Current rate: €0.05 / $0.05 per minute

Examples:
1 minute of audio → €0.05 / $0.05
60 minutes of audio → €3.00 / $3.00

Developer documentation

For full technical details, including request parameters, SDK usage examples, and supported language codes, refer to Lara's developer documentation:
https://developers.laratranslate.com/docs/translate-audio#audio-to-text-translation-and-transcription

This article is about:

• What audio-to-text translation is in Lara
• How it differs from audio-to-audio translation
• Supported input formats and file requirements
• Supported languages for input and output
• Customization options (translation memories, glossaries, style)
• How usage is calculated and billed