Applied Brain Research Releases the ABR SDK, Bringing Real-Time On-Device Voice Interfaces to Edge Applications

Niagara ASR and Nith TTS streaming models reach production release in a single SDK

WATERLOO, ON, Sept. 21, 2026 /PRNewswire/ -- Applied Brain Research (ABR) today announced general availability of the ABR SDK and the Niagara ASR and Nith TTS model families, a production toolkit for building real-time voice interfaces that run entirely on edge processor hardware.

The SDK provides streaming automatic speech recognition and streaming text-to-speech through a single API, so a developer can enable a responsive voice interface on the device itself. Because the latency-critical speech input and output processing runs locally, there is no speech AI dependency on a network, voice data does not leave the device, and the interface continues to operate when connectivity is unreliable.

"A voice interface is only usable if it answers immediately, which makes real-time performance the binding constraint for edge applications," said Kevin Conley, CEO of Applied Brain Research. "ABR's models and SDK are designed to deliver leading accuracy with the lowest latency on constrained edge hardware. Customers and hardware partners evaluating the SDK are consistently surprised by the response times and accuracy they measure."

Features

The current release provides a Python library over a stable C ABI, with C and Java bindings to follow in the coming weeks. A single SDK covers ABR's Niagara speech recognition and Nith speech synthesis families running simultaneously, starting with English, Spanish, Mandarin, Japanese and Korean.

Each model is distributed as a self-contained package holding the compiled library, the model weights and the configuration in one directory. Changing language or upgrading to newly trained weights requires only a change of the path. There is no inference backend to select and no cross-compilation step.

Two add-on capabilities are available to address brand- and domain-specific applications. For TTS, voice cloning generates a new synthesis voice from a short reference recording using consent-based training. For ASR and TTS, custom vocabulary allows a product's domain terms, proper nouns and product names to be recognized and pronounced correctly by both the recognition and synthesis models, without retraining.

Performance

ABR's Niagara streaming ASR models produce first text from the initial audio in as little as 115 milliseconds. The Nith streaming TTS models produce first audio from initial text in as little as 147 milliseconds, measured on embedded application-class CPUs.

All models operate faster than real time on every supported platform.

Supported platforms

Linux x86-64, Linux ARM64 and Android ARM64 are currently supported. Acceleration on integrated NPUs and DSPs is also available on certain partner platforms. Additional silicon targets and support for RTOS running on Cortex-M class microcontrollers with NPUs are targeted for release before year end.

Evaluation and availability

Documentation is available on ABR's documentation portal. The SDK and application packages are available from the ABR developer portal. Self-managed registration enables access to free evaluation of the full SDK and available models. Pilot and production deployment are available under commercial license from ABR.

ABR's open non-streaming (batch) speech recognition models appear on the Hugging Face Open ASR Leaderboard, where niagara-38m-batch.en is the most accurate model under 100 million parameters, at a mean word error rate of 9.69 percent as of the date of this release. niagara-19m-batch.en, one of the smallest on the leaderboard, outperforms models several times its size. Both can be tried in ABR's demo Space. The streaming models in the SDK use the same state space model architecture as these batch models and are trained on a superset of the same data.

About Applied Brain Research

Applied Brain Research develops efficient real-time speech and language AI that runs entirely on edge devices. Its Niagara ASR and Nith TTS model families deliver leading performance at a fraction of the size of comparable models. ABR derives significant efficiency and low latency from its state space model architectures, which originate from ABR's patented Legendre Memory Unit. ABR is based in Waterloo, Ontario, Canada. More at appliedbrainresearch.com.

SOURCE Applied Brain Research Inc.