Datasets
Explore rights-cleared datasets curated for real-world data needs. Browse sample collections across media, language, interaction, and structured formats.
AudioBengali Speech: Single & Multi-Speaker
Bengali speech with single-speaker narration and two-speaker conversations. Ground-truth transcripts and speaker metadata support ASR and voice-interface evaluation.
AudioHindi Speech: Contributor Recordings
Hindi speech recordings collected through a contributor recording app. Each clip is paired with validated written text and consistent audio metadata.
AudioSpanish Speech: Contributor Recordings
Spanish speech recordings collected through a contributor recording app. Each clip is paired with validated written text and consistent audio metadata.
AudioKorean Speech: Contributor Recordings
Korean speech recordings collected through a contributor recording app. Each clip is paired with validated written text and consistent audio metadata.
AudioUrdu Speech: Contributor Recordings
Urdu speech recordings collected through a contributor recording app. Each clip is paired with validated written text and consistent audio metadata.
AudioFrench Speech: Contributor Recordings
French speech recordings collected through a contributor recording app. Each clip is paired with validated written text and consistent audio metadata.
AudioBengali Speech: Read & Conversational
Bengali speech recordings captured across read prompts and natural conversational sessions. Each clip includes a validated transcript for ASR, voice interfaces, and language model alignment.
AudioTamil Speech: Read & Conversational
Tamil speech recordings with validated transcripts across diverse speaking styles, including read speech and conversational audio. Built for ASR, low-resource language research, and voice product development.
AudioTelugu Speech: Read & Conversational
Telugu speech audio covering scripted prompts, narration, and conversational samples. Each recording is paired with a transcript to support ASR evaluation, speech model tuning, and voice assistant workflows.
AudioChinese (Mandarin) Speech Samples
Chinese speech recordings in Mandarin from native speakers, paired with ground-truth transcripts and speaker metadata for speech recognition, dialect analysis, and voice interface evaluation.
AudioChinese (Cantonese) Speech Samples
Chinese speech recordings in Cantonese from native speakers, paired with ground-truth transcripts and speaker metadata for speech recognition, dialect analysis, and voice interface evaluation.
AudioChinese (Henan) Speech Samples
Henan Chinese speech recordings from native speakers, paired with ground-truth transcripts and speaker metadata for speech recognition, regional speech analysis, and voice interface evaluation.
VideoRobotic Arm: Pick-and-Place & Manipulation
Robotic-arm pick-and-place episodes with synchronized video, telemetry, camera information, and manifests for computer vision and robot-learning review.
VideoCultural Interviews: Public Conversations & Perspectives
Public interview videos featuring unscripted cultural conversations. Samples are structured for reviewing interview setting, conversational style, framing, and video quality before scoping broader coverage.
VideoEgocentric Video: Cleaning, Laundry & Car Wash
First-person task videos covering appliance cleaning, car wash and repair, interior work, and pet sitting, with task flow visible from the worker's view.
VideoHumanoid Robot Training: Egocentric Manipulation
Egocentric video of real-world manipulation tasks for robotics and embodied AI workflows, including household work, assembly operations, and human-robot collaboration scenarios.
VideoFirst-Person Task Video: Household & Maintenance Activities
Curated first-person videos across kitchen, outdoor, household, light maintenance, and fine-motor tasks, with clear hand-object interaction in each sample.
VideoWarehouse: Kitting, Pick-and-Place & Packaging
Human workers performing kitting, pick-and-place, and packaging tasks in a simulated warehouse setting for logistics and robotics evaluation.
VideoBehavioral Dataset: Multimodal Video, Depth & Actions
Human-teleoperated demonstrations of everyday household tasks with synchronized RGB video, depth, and instance segmentation for embodied AI workflows.
VideoComputer Use: Agent Operating a Desktop Interface
Agent-view recordings of desktop interaction, including cursor movement, GUI actions, and on-screen task execution for computer-use agents.
3DMotion Capture: Actor Performances & 3D Objects
3D scanned characters and objects paired with professional motion-capture performances for digital human and animation pipelines.
PairsQ&A Pairs: Physics, Chemistry, Math & Biology
Verified Q&A pairs across Physics, Chemistry, Math, and Biology, aligned with competitive Indian engineering exams (JEE, NEET, CET). Includes textual and image-based content tagged by difficulty, curriculum, and type.
PairsCode I/O Pairs: Prompt-Driven Few-Shot Workflows
Developer workflow pairs with aligned prompts, code, tests, error traces, and corrected outputs for code generation and LLM evaluation.
Looking for specific data?
Tell us what you need, including format, coverage, language, region, or licensing requirements.
Request a dataset