Turn-milling is a kind of machining process which widely applied in industries with large-sized workpieces. It is because turn-milling provides advantages for machining large-diameter mechanical parts with high speed, reducing cutting temperature due to the chips being broken, which in turn decreases tool wear. However, it needs to monitor the turn-milling process for preventing the onset of chatter during operation. It is due to the chatter becoming a limitation to productivity, part quality, accelerates tool wear, and machine-tool damage. One of the ways for turn-milling process monitoring is by vibration analysis. Acquired vibrations in the machining process are generally backgrounded by noises and the conventional filtered tools may have defiance for reducing them. It is significant to find signal processing tools for denoising noisy signals before further analysis. This paper presents the utilization of the empirical mode decomposition (EMD) method as an efficient and adaptive noise filter. The Short-Time Fourier Transform (STFT) improvement using EMD is then used for monitoring turn-milling process conditions in the energytime-frequency domain. The results showed that the reconstructed signal was quite impressive compared to the raw signal and the oscillation of the filtered signal was clearer than the raw signal. The improvement of the filtered signals was proved by the kurtosis index and spectral kurtosis. The kurtosis index had been improved 10 – 27 times more than raw signals. The improved STFT using EMD showed a significant spectrum with high resolution compared to conventional STFT. The energy density could be observed clearly in the machining characteristic frequencies with an improvement of about 10-100 times larger. The proposed method is therefore effectively applied to monitor the turn-milling condition.