Automatic Speech Recognition (ASR) provides a powerful tool for preserving endangered languages. However, the effectiveness of ASR systems relies on the availability of high-quality speech data, which is often scarce for many languages, including Balti. This study presents the first documented effort to create a spoken Balti corpus, introducing this endangered language to the speech research community and advancing its preservation. We compile the first-ever Balti speech dataset, comprising 10,394 voice recordings of 473 isolated words (primarily nouns) spoken by 39 native speakers, paired with transcriptions. Using this corpus, we evaluate ASR performance through statistical (GMM-HMM) and neural (TDNN) approaches, along with fine-tuned pretrained Whisper models. The experiments achieve word error rates (%WER) of 20.8 on speaker-independent (SI), 13.4 on speaker-adaptive Training (SAT), and 8.72 on time delay neural network (TDNN). Fully fine-tuning yields further significant improvements, with %WERs of 12.98 on Tiny, 10.29 on Base, 7.72 on Small, and 6.04 on Medium Whisper models, respectively. Despite the limited size of the dataset, constraints of isolated words, and unknown language in the multilingual models, these results demonstrate the feasibility of it in the recognition task, and also for the preservation of Balti, establishing a critical foundation for future research.