--- title: "Portuguese Speech Recognition Dataset" description: "The dataset contains 10+ hours of annotated telephone dialogues from 20+ native Portuguese speakers, offering detailed audio recordings, transcriptions, and speaker metadata to train speech…" url: "https://unidata.pro/datasets/portuguese-speech-recognition-dataset/" date_modified: "2025-12-10T16:48:50+03:00" language: "en-US" --- The dataset contains 10+ hours of annotated telephone dialogues from 20+ native Portuguese speakers, offering detailed audio recordings, transcriptions, and speaker metadata to train speech recognition systems, NLP models, and machine learning applications with diverse Portuguese speech datasets ## Dataset Structure ### The Numbers Section **Numbered list:** - **Number:** 10+ — **Text:** Hours - **Number:** 20+ — **Text:** Speakers ### Tooltips Section **Tooltip items:** - **Name:** NLP - **Name:** LLM - **Name:** Machine Learning - **Name:** Audio Processing - **Name:** ASR - **Name:** Voice Recognition ### Dataset Information **Table with data:** | Characteristic | Data | | --- | --- | | Description | Audio of telephone dialogues in Portuguese for training NLP models in real-world conversational scenarios. | | Data types | Audio | | Tasks | Speech recognition, NLP | | Country | Portugal(PRT) | | Hours of telephone dialogue | 10 | | Number of speakers | 20 | | Labeling | Annotation (ID, Language, Format, Minutes) | | Recording device | Android smartphone, iPhone | **Media Slider:** - **Video on Slayder:** - **Video on Slayder:** **Link to the sample:** [Download sample](https://drive.google.com/drive/folders/1LMDK_mluGnw-ZGe67JVkUfXv-b254sfv?usp=sharing) ### Technical Specifications **Table with data:** | Characteristic | Data | | --- | --- | | Audio Format | WAV, M4A, MP3 | | Recording condition | Low background noise (indoor) | | Duration | Mean =11 min | **Source and data collection methodology:** Source and collection methodology: Data was collected via crowdsourcing platforms ### Dataset Use Cases - Slider **Industry Cards:** - **Industry:** Call Centers and Customer Service — **Title:** Improving Speech Recognition in Telephone Conversations — **Text:** Portuguese Speech Recognition Dataset provides telephone dialogues dataset samples collected from native speakers with various accents. These audio recordings and transcripts help train recognition systems for call centers, enabling accurate transcribing of speech, detecting customer intent, and supporting language processing tasks with higher accuracy in real-world commercial use cases. - **Industry:** Healthcare and Telemedicine — **Title:** Enhancing Voice Technology for Remote Consultations — **Text:** This Portuguese audio dataset supports medical applications by offering speech samples suited for healthcare communication systems. Automatic speech recognition and language processing models trained on this diverse dataset achieve better accuracy in understanding spoken words, helping doctors and patients interact seamlessly in telemedicine platforms using natural language. - **Industry:** AI and Machine Learning Development — **Title:** Training Models for Speech Processing Tasks — **Text:** The dataset consists of high-quality Portuguese telephone dialogues and audio samples used for machine learning and deep learning projects. With diverse accents and natural speech signals, it improves recognition technology, builds multilingual speech models, and supports custom datasets for voice assistants, transcription services, and automatic speech translation systems. - **Industry:** Education and Language Technology — **Title:** Building Tools for Portuguese Language Learning — **Text:** This dataset enables the creation of speech processing applications for language learning platforms. By using annotated audio files with accurate transcriptions, learning models can better process natural speech, recognize voice patterns, and adapt to various accents, helping learners gain proficiency in Portuguese through interactive voice technology. ### Fact **FAQs Heading:** FAQs **List of Questions:** - **Question:** What types of annotations are provided? — **Answer:** The dataset includes fully transcribed dialogues along with metadata annotations for speaker ID, language, formats, minutes. These labels are essential for building speech recognition models with high accuracy. - **Question:** Is the dataset suitable for commercial use? — **Answer:** Yes, the dataset is licensed for both research and commercial usage. It can be applied in customer service systems, voice recognition software, speech-to-text solutions, and other language processing technologies. - **Question:** Why is native speaker audio valuable for training machine learning models? — **Answer:** Native speaker recordings provide authentic examples of vocabulary, pronunciation, and speaking styles that improve model understanding of a language. This helps reduce recognition errors and creates more reliable speech technologies for Portuguese-speaking users. - **Question:** What should I consider before buying this dataset? — **Answer:** When purchasing it, check the audio format, sampling rate, and the number of native Portuguese speakers included. Make sure the annotations and speech samples match your goals for speech recognition, NLP models, or machine learning projects. - **Question:** Do Unidata datasets follow GDPR or other data privacy regulations? — **Answer:** Yes. All Unidata datasets are curated in compliance with GDPR and international privacy laws. The audio recordings were collected only from legally permissible sources to ensure ethical data collection and safe commercial usage. - **Question:** How long does it take to receive the dataset? — **Answer:** After submitting a request, Unidata will review the details and complete the necessary documentation with you. Once the agreement is signed and payment is received, the dataset will be delivered within 3-10 days. - **Question:** Is it unique data? — **Answer:** Yes. This dataset consists of telephone speech collected directly from real participants through Unidata’s partner, making it unique and not available in open-source repositories. - **Question:** Is this a real-world dataset or synthetic data? — **Answer:** This is a real-world dataset. Portuguese Speech Recognition Dataset was collected from native speakers using Android and iPhone devices in natural recording conditions with low background noise. [Full list of this site's AI-readable pages](https://unidata.pro/llms.txt)