IBM has released Granite 4.0 1B Speech, a compact, multilingual speech model designed for enterprise applications on resource-constrained devices. As the latest addition to the Granite Speech collection, this model specializes in automatic speech recognition and bidirectional speech translation. With only half the parameters of its predecessor, granite-speech-3.3-2b, the new model delivers higher English transcription accuracy, faster inference speeds through speculative decoding, and support for six languages: English, French, German, Spanish, Portuguese, and Japanese. Two significant enhancements in this release address frequent community requests. The model now includes Japanese automatic speech recognition capabilities and introduces keyword list biasing, which improves the accuracy of recognizing specific names and acronyms. Despite its small size, Granite 4.0 1B Speech has achieved competitive results on standard benchmarks. It recently ranked number one on the OpenASR leaderboard, outperforming many larger open speech recognition systems. Performance is measured using Word Error Rate, where a lower score indicates higher accuracy. The model demonstrates strong results across multiple datasets while utilizing significantly fewer parameters than comparable models. All Granite models, including this new release, are available under the Apache 2.0 license and feature native support in transformers and vLLM environments. IBM evaluated the model across a wide range of standard speech recognition and translation tasks. The results show that the model performs as well as, or better than, much larger systems despite its minimal parameter count. Detailed evaluation results, architecture specifications, training data information, and usage examples are available in the official model card. For production deployments requiring additional risk detection, IBM recommends pairing the model with Granite Guardian. The release marks a significant step in making high-performance speech AI accessible on edge devices without compromising accuracy or speed.