AERO

AERO (AI-Enhancements Responsive to Orality) addresses the time- and resource-consuming hurdles faced by Oral Bible Translation teams through advanced AI audio processing. However, it can be used for editing audio for other purposes, as well.

Noise Removal: Strips out background noise from suboptimal recording environments like chickens crowing, an air conditioner running, or other people in the background of an OBT workshop.

Audio Enhancement: Goes a step further, regenerating clearer, more intelligible speech from a degraded recording, which is useful when the original is muffled, distant, or distorted rather than simply noisy.

Voice Conversion: Disguises a speaker's voice while preserving the clarity of the language, even making it sound like the speaker is a different gender [to add new voices in a dramatized recording]. Users confirm they have the target speaker's permission before converting.

Transcription: Offers speech-to-text in over 1,600 languages, reducing the manual transcription workload for field teams. AERO recommends the best-performing model for the language you select. Custom models are available on request in other languages if data is available. In languages without a writing system, a sister-language may be used with a writing system that isn't completely foreign to the language community.

Text to Speech: Generates natural-sounding speech from text in a wide range of languages.

Find & Replace: Finds words or phrases in audio files and replaces them without noticeable transitions in the same voice and without re-recording the whole passage. Replacement audio can be uploaded, recorded, or generated from text.

AERO is accessible as a web application and via API, and features integration with SIL's Audio Project Manager (APM).

Learn more about AERO and other SIL AI projects at ai.sil.org.