A new method was developed for the separation of source and transfer function characteristics of speech sounds, with an aim of utilizing it to "flexible" speech synthesis. The method is based on representing source waveform by an HMM, and transfer function by the AR process (AR-HMM model). As compared to methods based on ARX model, where a parametric representation is assumed for source waveform, a better separation is possible. By introducing a process of recursively deleting real poles of AR filters, which represent source waveform features, and including them intoHMMsource waveform, the resulting AR filters may correctly represent transfer function fea- tures. Experiments were conducted for Japanese vowel sounds in continuous speech, and the results were compared with those by conventional LP analysis and ARHMM model analysis without recursive process. After representing obtained source and transfer function features respectively as DFT cepstrum and LPC cepstrum, variations of cepstrum parameters for each vowel sound were compared for the three analysis methods. The smallest variations were obtained by the proposed method, indicating that the proposed method can separate source and transfer function features well, and, thus, has potential ability of generating good quality of speech when applied to "flexible" speech synthesis.