Volatility is a key model parameter of the financial market. It controls the risk associated to the stochastic dynamics of the asset prices and also affects the rational price of derivative products. There are empirical evidences that the volatility is also a stochastic process, underlined to the price one. Therefore, the volatility is not directly observed and must be estimated, constituting a hidden stochastic process. In this work, we consider an estimate for the daily volatility of the BOVESPA index, computed from the intraday database. We perform a descriptive statistical analysis of the resulting time series, obtaining the probability density function, moments and correlations. We compare the empirical outcomes with the theoretical forecasts of many stochastic volatility models. We consider the class of Itô-Langevin equations composed by a mean reverting process and a generalized diffusive Wiener process with multiplicative and/or additive noise components. From this analysis, we propose a model that describes the volatility fluctuations of the Brazilian stock market.