This research work presents a new valid dataset for Punjabi called the Punjabi Emotional Speech Database (PEMO) which has been developed to assess the ability to recognize emotions in speech by both computers and humans. The PEMO includes speech samples from about 60 speakers with an age range between 20 and 45 years, for four fundamental emotions, including anger, sad, happy and neutral. In order to create the data, Punjabi films are retrieved from different multimedia websites such as YouTube. The movies are processed and transformed into utterances with software called PRAAT. The database contains 22,000 natural utterances. This is equivalent to 12 hours and 35 min of speech information taken from online Punjabi movies and web series. Three annotators categorize the emotional content of the utterances. The common label that is labelled by all annotators becomes the final label for the utterance. The annotators have a thorough knowledge of Punjabi Language. The data is used to determine the expression of emotions in speech in the Punjabi Language.