ON-DEVICE SPEECH RECOGNITION

Offline streaming transcription

Every voice.
Into words.

The ELEJO offline speech recognition SDK transcribes microphone input or an audio stream on the device. Once the matching model and license are in place, transcription works offline.

Android / iOSC++ SDKLocal models
Explore in 3D
Voice to textOn your device

We plan to integrate an on-device model into our next generation of devices.

Next, let’s review the local deployment plan.

The formula behind it
s(x,t) = Σ aₖ sin(kx − ωₖt)

This curve shows how waves at different frequencies add together.

01 / FROM SOUND TO TEXT

Bring voice into your product.
Keep processing local.

Connect audio capture, recognition results, and domain vocabulary into a complete experience for recordings, meetings, and voice interactions.

01
INPUT

Receive audio

Capture a microphone or audio stream and prepare the sample rate and format required by the SDK.

02
ON DEVICE

Recognize locally

Run the engine on the target device, process audio continuously, and configure vocabulary for your domain.

03
OUTPUT

Display text

Receive recognition results and connect them to your notes, captions, or interaction interface.

02 / YOUR WORDS, YOUR CONTEXT

Specialist language.
Familiar context.

Brand names, industry terms, and product vocabulary. Add key expressions to local vocabulary settings, then verify recognition with recordings from real use cases.

Read integration docs
Brand nameElejo AI01
Technical termOn-device model02
Business vocabularyProduct solution03
READY FOR YOUR DEVICE

Start with your audio.

Bring your platform, audio samples and vocabulary. Build around the way your product is used.

  1. 01Confirm audio and platform
  2. 02Choose SDKs and models
  3. 03Test real recordings

Languages, audio formats, resource use, and latency depend on the released SDK compatibility notes and test reports.