iFlytek has announced that SparkAudio 1.0, its Spark speech foundation large model, is now available as an API service on the iFlytek Open Platform, enabling developers to integrate and call it directly. The model accepts both text and speech as input and produces text output, supporting tasks such as speech transcription, multilingual translation, multi-dialect recognition, environmental sound understanding, speaker separation, sentiment analysis, and complex audio question answering. It overcomes the shortcomings of traditional cascaded audio processing pipelines—namely information loss and fragmented workflows—by understanding speech directly while simultaneously capturing the semantic, emotional, and scene information conveyed through human voices, environmental sounds, music, and audio. Compared with its predecessor, the new model brings upgrades on three fronts. First, it supports 99 languages and 202 dialects, enabling multilingual speech recognition and translation. Second, it has evolved from speech recognition to multidimensional audio understanding: it can interpret the deeper meaning behind speech and is well suited to scenarios such as intelligent customer service, automotive cockpits, cross-language meetings, medical records, and content moderation. Third, it is trained entirely on domestic computing power, giving developers and industry applications an autonomous and controllable foundation for speech intelligence.