# Qualcomm Dragonwing Documentation: Ubuntu ## Ubuntu ### Ubuntu Overview - [Ubuntu overview](https://dragonwingdocs.qualcomm.com/Ubuntu/ubuntu-overview.md): [Ubuntu on Qualcomm® IoT platforms](https://www.qualcomm.com/developer/software/ubuntu-on-qualcomm-iot-platforms) simplifies access to Qualcomm’s advanced AI capabilities, empowering developers and enterprises to innovate with confidence. Run Ubuntu on Qualcomm IoT platforms to accelerate the deploy… - [Platform architecture](https://dragonwingdocs.qualcomm.com/Ubuntu/ubuntu-platform-architecture.md): Understand the main layers of the Ubuntu on Qualcomm® IoT platforms software stack. - [Supported hardware](https://dragonwingdocs.qualcomm.com/Ubuntu/ubuntu-supported-hardware.md): Compare the Dragonwing hardware covered by the Ubuntu documentation. - [Development workflows](https://dragonwingdocs.qualcomm.com/Ubuntu/ubuntu-development-workflows.md): Map common Ubuntu development tasks to the documentation sections that support them. - [Tools and acceleration](https://dragonwingdocs.qualcomm.com/Ubuntu/ubuntu-tools-and-acceleration.md): Review Ubuntu tools, AI runtimes, multimedia paths, and developer utilities for Dragonwing platforms. ### IQ-9075 EVK #### IQ-9075 EVK - [Device overview](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/device-overview.md): Hardware overview of the Dragonwing IQ-9075 EVK including SoC features, peripherals, and board layout. - [Set up the device](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/set-up-the-device.md): After flashing Ubuntu, connect the serial console, verify the version, and bring up network, SSH, and display on the IQ-9075 EVK. - [Install Required Software Packages](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/Install_required_software_packages.md) - [Development device regulatory notice](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/Development_device_regulatory_notice.md) ##### Update software - [Overview](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/update-software/overview.md): Install or update Ubuntu on the IQ-9075 EVK, and decide whether you need to flash. - [Flash Ubuntu using Qualcomm® Launcher](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/update-software/flash-using-qualcomm-launcher.md): Use the Qualcomm Launcher GUI tool to download and flash Ubuntu onto the IQ-9075 EVK from any host OS. - [Flash Ubuntu using QDL](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/update-software/flash-using-qdl.md): Use the Qualcomm Device Loader (QDL) command-line tool to flash Ubuntu onto the IQ-9075 EVK from Ubuntu, Windows, or macOS hosts. - [Upgrade Ubuntu image](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/update-software/upgrade-ubuntu.md) ##### Peripherals and interfaces - [Fan](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/Fan.md) - [DIP Switches, Ports and Interfaces](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/Switches.md) - [40 Pin Low Speed Connector](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/Low-Speed-Connectors.md) - [Mezzanine Boards](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/accessory-cards.md) - [Avl](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/avl.md) ###### Multimedia - [Camera](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/Camera.md) - [Audio](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/Audio.md) - [Display](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/Display.md) - [Sensors](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/Sensors.md) ###### Network - [WiFi and Bluetooth](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/WiFi-BT.md) - [Ethernet](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/Ethernet.md) ###### Interfaces - [Overview](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/peripheral_interface_overview.md) - [USB](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/USB.md) - [PCIe](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/PCIe.md) - [UART](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/UART.md) - [I2C](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/I2C.md) - [I3C](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/I3C.md) - [SPI](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/SPI.md) - [LED](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/LED.md) ###### Memory - [LPDDR Memory](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/LPDDR_memory.md) - [Memory and Storage](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/peripherals-interfaces/memory_storage.md) ##### Sample applications - [Overview](https://dragonwingdocs.qualcomm.com/Ubuntu/sample-applications/overview.md) - [Prebuilt sample applications](https://dragonwingdocs.qualcomm.com/Ubuntu/sample-applications/prebuilt-sample-applications.md) - [Qualcomm® Qdemo](https://dragonwingdocs.qualcomm.com/Ubuntu/sample-applications/qdemo.md) - [Configure sample applications](https://dragonwingdocs.qualcomm.com/Ubuntu/sample-applications/config_applications.md) - [Customize existing sample applications](https://dragonwingdocs.qualcomm.com/Ubuntu/sample-applications/customize_existing_sample_application.md) ##### Running / building AI models - [AI Workflow Overview](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/ai-workflows_overview.md): Overview of supported AI workflows for training, converting, and deploying models on Qualcomm Dragonwing devices. - [Edge Impulse](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/edge-impulse.md): Build and deploy edge AI models on Dragonwing devices using the Edge Impulse platform with full NPU acceleration. - [Qualcomm AI Hub](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/ai-hub.md): Deploy pretrained AI models from Qualcomm AI Hub to the NPU of your Dragonwing board using LiteRT or ONNX Runtime. - [Convert TensorFlow models](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/tensorflow.md): Quantize and convert TensorFlow/Keras models to TFLite format so they can run on the NPU via LiteRT. - [Run PyTorch models on NPU](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/run-py-torch-models-on-npu.md) - [Run LiteRT / TFLite models](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/lite-rt.md): Run quantized LiteRT models on the NPU of Dragonwing devices using AI Engine Direct delegates in Python or C++. - [Run ONNX models](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/onnxruntime.md): Run ONNX models on the NPU using ONNX Runtime with AI Engine Direct on Dragonwing devices. - [Run context binaries (.bin/.dlc)](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/context-binaries.md) - [LLMs/VLMs using Llama.cpp](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/llama-cpp.md): Run large language and vision-language models locally on Dragonwing devices using llama.cpp, with CPU, GPU (OpenCL), or NPU (Hexagon HTP) backends. - [Whisper](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/whisper.md) - [Qualcomm® Intelligent Multimedia (IM) SDK](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/im-sdk.md): Build zero-copy AI vision pipelines using GStreamer plugins that run on the GPU and NPU of Dragonwing devices. ###### LLM/VLM using Qualcomm OpenAI container - [Running the LLM/VLM container](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/llm-vlm-qcom-setup.md) - [Using with Langchain](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/llm-vlm-qcom-langchain.md) - [Running on a specific NPU (QCS9075)](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/llm-vlm-qcom-npuconfig.md) ###### API ###### Chat - [Creates a chat conversation.](https://dragonwingdocs.qualcomm.com/api-reference/chat/creates-a-chat-conversation.md) - [Delete a stored chat completion](https://dragonwingdocs.qualcomm.com/api-reference/chat/delete-a-stored-chat-completion.md): Deletes a chat completion by ID, cleaning up all associated resources. - [Cancel an active chat completion](https://dragonwingdocs.qualcomm.com/api-reference/chat/cancel-an-active-chat-completion.md): Cancels an actively executing chat completion by terminating the subprocess and rolling back session state. ###### Health - [Check the health of microservice](https://dragonwingdocs.qualcomm.com/api-reference/health/check-the-health-of-microservice.md): Health check for container orchestration (Docker Swarm / Kubernetes). Returns HTTP 200 when healthy, HTTP 503 when unhealthy. Unhealthy conditions: any model has >= 3 consecutive inference failures (DSP/FastRPC fault), or available system memory < 100 MB. ###### Ping - [Check if server is accessible to the client](https://dragonwingdocs.qualcomm.com/api-reference/ping/check-if-server-is-accessible-to-the-client.md): Ping the server to check it's alive and well. Ping will return Pong if you provide the correct BearerToken ###### Metrics - [Get system and model inference metrics](https://dragonwingdocs.qualcomm.com/api-reference/metrics/get-system-and-model-inference-metrics.md): Returns a unified snapshot of live system resource metrics (CPU, memory) and per-model moving averages for all pipeline stages (last 100 requests). ###### Models - [Lists the currently available models](https://dragonwingdocs.qualcomm.com/api-reference/models/lists-the-currently-available-models.md): Lists the currently available models, and provides basic information about each one such as the owner and availability. ###### Voice (ASR/TTS/translation) - [Running the Audio Analytics container](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/audio-analytics-overview.md): Deploy and run the Audio Analytics container for on-device ASR, text translation, and TTS. - [Speech recognition (ASR)](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/asr-container.md): Transcribe audio to text using on-device ASR models running on the Qualcomm NPU. - [Text-to-speech (TTS)](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/tts-container.md): Synthesize speech from text using on-device TTS models running on the Qualcomm NPU. - [Language translation](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/language-translation.md): Translate text between languages using neural machine translation models running on the Qualcomm NPU. ###### API ###### Health - [Check service health](https://dragonwingdocs.qualcomm.com/api-reference/health/check-service-health.md): Returns the current health status, uptime, active session count, and version. ###### Transcriptions - [List available ASR models](https://dragonwingdocs.qualcomm.com/api-reference/transcriptions/list-available-asr-models.md): Returns available ASR models with capabilities and supported parameters. - [Transcribe audio to text](https://dragonwingdocs.qualcomm.com/api-reference/transcriptions/transcribe-audio-to-text.md): Transcribes audio input into text. Supports three modes: - [Close active transcription session](https://dragonwingdocs.qualcomm.com/api-reference/transcriptions/close-active-transcription-session.md): Ends the current transcription session and releases server resources. - [Flush buffered audio and force immediate transcription](https://dragonwingdocs.qualcomm.com/api-reference/transcriptions/flush-buffered-audio-and-force-immediate-transcription.md): Forces the ASR engine to process any audio currently buffered in the VAD/silence-detection pipeline and emit a transcription result immediately, without waiting for the next speech-end event. ###### Translations - [List available translation models](https://dragonwingdocs.qualcomm.com/api-reference/translations/list-available-translation-models.md): Returns available translation models with supported language pairs and parameters. - [Translate text between languages](https://dragonwingdocs.qualcomm.com/api-reference/translations/translate-text-between-languages.md): Translates text from a source language to a target language. - [Close active translation session](https://dragonwingdocs.qualcomm.com/api-reference/translations/close-active-translation-session.md): Ends the current translation session and releases server resources. ###### TTS - [List available TTS models and voices](https://dragonwingdocs.qualcomm.com/api-reference/tts/list-available-tts-models-and-voices.md): Returns available TTS models with voices, capabilities, and supported sample rates. Filter by language using the `language` query parameter. - [Synthesize speech from text](https://dragonwingdocs.qualcomm.com/api-reference/tts/synthesize-speech-from-text.md): Converts text into spoken audio using text-to-speech synthesis. - [Close active text-to-speech session](https://dragonwingdocs.qualcomm.com/api-reference/tts/close-active-text-to-speech-session.md): Ends the current TTS session and releases server resources. - [Cancel an active text-to-speech playback](https://dragonwingdocs.qualcomm.com/api-reference/tts/cancel-an-active-text-to-speech-playback.md): Immediately stops an in-progress TTS synthesis or playback. The session remains open. ###### WebSocket - [Upgrade to WebSocket for real-time streaming](https://dragonwingdocs.qualcomm.com/api-reference/websocket/upgrade-to-websocket-for-real-time-streaming.md): Upgrades the HTTP connection to a WebSocket for real-time audio streaming. ##### Robotics workflows - [Robotics Workflow Overview](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/robotics-workflows_overview.md) - [Software Setup](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/software-setup.md) - [Depth Estimation on the NPU](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/npu-workflows.md): Build a ROS 2 depth estimation node that runs a quantized TFLite model on the Hexagon HTP NPU via the QNN delegate. ###### QRB ROS ecosystem - [QRB ROS Ecosystem](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/qrb-ros-overview.md) - [QRB ROS Samples](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/qrb-ros-samples.md) - [QRB ROS Transport](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/qrb-ros-transport.md) - [QRB ROS NN Inference](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/qrb-ros-nn-inference.md) - [QRB ROS Simulation](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/qrb-ros-simulation.md) - [QRB ROS Benchmark](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/qrb-ros-benchmark.md) - [ROS 2 DDS Config Optimizer](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/qrb-ros-dds-optimizer.md) ##### Tools - [Develop with Visual Studio Code via SSH](https://dragonwingdocs.qualcomm.com/Ubuntu/tools/vscode-ssh-setup.md) - [Docker](https://dragonwingdocs.qualcomm.com/Ubuntu/tools/docker.md) - [Profiling models](https://dragonwingdocs.qualcomm.com/Ubuntu/tools/profiling-models.md) - [libqcperf](https://dragonwingdocs.qualcomm.com/Ubuntu/tools/libqcperf.md): A lightweight C library for real-time performance monitoring on Qualcomm chipsets, with pluggable backends for CPU, NPU, and power/thermal metrics. ##### Legal - [Legal Information](https://dragonwingdocs.qualcomm.com/Ubuntu/Legal_notice.md) ### IQ-8275 EVK #### IQ-8275 EVK - [Device overview](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/device-overview.md): Hardware overview of the Dragonwing IQ-8275 EVK including SoC features, peripherals, and board layout. - [Set up the device](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/set-up-the-device.md): After flashing Ubuntu, connect the serial console, verify the version, and bring up network, SSH, and display on the IQ-8275 EVK. - [Install Required Software Packages](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/Install_required_software_packages.md) - [Development device regulatory notice](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/Development_device_regulatory_notice.md) ##### Update software - [Overview](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/update-software/overview.md): Install or update Ubuntu on the IQ-8275 EVK, and decide whether you need to flash. - [Flash Ubuntu using Qualcomm® Launcher](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/update-software/flash-using-qualcomm-launcher.md): Use the Qualcomm Launcher GUI tool to download and flash Ubuntu onto the IQ-8275 EVK from any host OS. - [Flash Ubuntu using QDL](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/update-software/flash-using-qdl.md): Use the Qualcomm Device Loader (QDL) command-line tool to flash Ubuntu onto the IQ-8275 EVK from Ubuntu, Windows, or macOS hosts. - [Upgrade Ubuntu image](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/update-software/upgrade-ubuntu.md) ##### Peripherals and interfaces - [Fan](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/Fan.md) - [DIP Switches, Ports and Interfaces](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/Switches.md) - [40 Pin Low Speed Connector](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/Low-Speed-Connectors.md) - [Mezzanine Boards](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/accessory-cards.md) - [Avl](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq9075-evk/avl.md) ###### Multimedia - [Camera](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/Camera.md) - [Audio](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/Audio.md) - [Display](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/Display.md) - [Sensors](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/Sensors.md) ###### Network - [WiFi and Bluetooth](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/WiFi-BT.md) - [Ethernet](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/Ethernet.md) ###### Interfaces - [Overview](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/peripheral_interface_overview.md) - [USB](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/USB.md) - [PCIe](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/PCIe.md) - [UART](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/UART.md) - [I2C](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/I2C.md) - [I3C](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/I3C.md) - [SPI](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/SPI.md) - [LED](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/LED.md) ###### Memory - [LPDDR Memory](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/LPDDR_memory.md) - [Memory and Storage](https://dragonwingdocs.qualcomm.com/Ubuntu/devices/iq8275-evk/peripherals-interfaces/memory_storage.md) ##### Sample applications - [Overview](https://dragonwingdocs.qualcomm.com/Ubuntu/sample-applications/overview.md) - [Prebuilt sample applications](https://dragonwingdocs.qualcomm.com/Ubuntu/sample-applications/prebuilt-sample-applications.md) - [Qualcomm® Qdemo](https://dragonwingdocs.qualcomm.com/Ubuntu/sample-applications/qdemo.md) - [Configure sample applications](https://dragonwingdocs.qualcomm.com/Ubuntu/sample-applications/config_applications.md) - [Customize existing sample applications](https://dragonwingdocs.qualcomm.com/Ubuntu/sample-applications/customize_existing_sample_application.md) ##### Running / building AI models - [AI Workflow Overview](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/ai-workflows_overview.md): Overview of supported AI workflows for training, converting, and deploying models on Qualcomm Dragonwing devices. - [Edge Impulse](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/edge-impulse.md): Build and deploy edge AI models on Dragonwing devices using the Edge Impulse platform with full NPU acceleration. - [Qualcomm AI Hub](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/ai-hub.md): Deploy pretrained AI models from Qualcomm AI Hub to the NPU of your Dragonwing board using LiteRT or ONNX Runtime. - [Convert TensorFlow models](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/tensorflow.md): Quantize and convert TensorFlow/Keras models to TFLite format so they can run on the NPU via LiteRT. - [Run PyTorch models on NPU](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/run-py-torch-models-on-npu.md) - [Run LiteRT / TFLite models](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/lite-rt.md): Run quantized LiteRT models on the NPU of Dragonwing devices using AI Engine Direct delegates in Python or C++. - [Run ONNX models](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/onnxruntime.md): Run ONNX models on the NPU using ONNX Runtime with AI Engine Direct on Dragonwing devices. - [Run context binaries (.bin/.dlc)](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/context-binaries.md) - [LLMs/VLMs using Llama.cpp](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/llama-cpp.md): Run large language and vision-language models locally on Dragonwing devices using llama.cpp, with CPU, GPU (OpenCL), or NPU (Hexagon HTP) backends. - [Whisper](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/whisper.md) - [Qualcomm® Intelligent Multimedia (IM) SDK](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/im-sdk.md): Build zero-copy AI vision pipelines using GStreamer plugins that run on the GPU and NPU of Dragonwing devices. ###### LLM/VLM using Qualcomm OpenAI container - [Running the LLM/VLM container](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/llm-vlm-qcom-setup.md) - [Using with Langchain](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/llm-vlm-qcom-langchain.md) - [Running on a specific NPU (QCS9075)](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/llm-vlm-qcom-npuconfig.md) ###### API ###### Chat - [Creates a chat conversation.](https://dragonwingdocs.qualcomm.com/api-reference/chat/creates-a-chat-conversation.md) - [Delete a stored chat completion](https://dragonwingdocs.qualcomm.com/api-reference/chat/delete-a-stored-chat-completion.md): Deletes a chat completion by ID, cleaning up all associated resources. - [Cancel an active chat completion](https://dragonwingdocs.qualcomm.com/api-reference/chat/cancel-an-active-chat-completion.md): Cancels an actively executing chat completion by terminating the subprocess and rolling back session state. ###### Health - [Check the health of microservice](https://dragonwingdocs.qualcomm.com/api-reference/health/check-the-health-of-microservice.md): Health check for container orchestration (Docker Swarm / Kubernetes). Returns HTTP 200 when healthy, HTTP 503 when unhealthy. Unhealthy conditions: any model has >= 3 consecutive inference failures (DSP/FastRPC fault), or available system memory < 100 MB. ###### Ping - [Check if server is accessible to the client](https://dragonwingdocs.qualcomm.com/api-reference/ping/check-if-server-is-accessible-to-the-client.md): Ping the server to check it's alive and well. Ping will return Pong if you provide the correct BearerToken ###### Metrics - [Get system and model inference metrics](https://dragonwingdocs.qualcomm.com/api-reference/metrics/get-system-and-model-inference-metrics.md): Returns a unified snapshot of live system resource metrics (CPU, memory) and per-model moving averages for all pipeline stages (last 100 requests). ###### Models - [Lists the currently available models](https://dragonwingdocs.qualcomm.com/api-reference/models/lists-the-currently-available-models.md): Lists the currently available models, and provides basic information about each one such as the owner and availability. ###### Voice (ASR/TTS/translation) - [Running the Audio Analytics container](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/audio-analytics-overview.md): Deploy and run the Audio Analytics container for on-device ASR, text translation, and TTS. - [Speech recognition (ASR)](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/asr-container.md): Transcribe audio to text using on-device ASR models running on the Qualcomm NPU. - [Text-to-speech (TTS)](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/tts-container.md): Synthesize speech from text using on-device TTS models running on the Qualcomm NPU. - [Language translation](https://dragonwingdocs.qualcomm.com/Ubuntu/ai-workflows/language-translation.md): Translate text between languages using neural machine translation models running on the Qualcomm NPU. ###### API ###### Health - [Check service health](https://dragonwingdocs.qualcomm.com/api-reference/health/check-service-health.md): Returns the current health status, uptime, active session count, and version. ###### Transcriptions - [List available ASR models](https://dragonwingdocs.qualcomm.com/api-reference/transcriptions/list-available-asr-models.md): Returns available ASR models with capabilities and supported parameters. - [Transcribe audio to text](https://dragonwingdocs.qualcomm.com/api-reference/transcriptions/transcribe-audio-to-text.md): Transcribes audio input into text. Supports three modes: - [Close active transcription session](https://dragonwingdocs.qualcomm.com/api-reference/transcriptions/close-active-transcription-session.md): Ends the current transcription session and releases server resources. - [Flush buffered audio and force immediate transcription](https://dragonwingdocs.qualcomm.com/api-reference/transcriptions/flush-buffered-audio-and-force-immediate-transcription.md): Forces the ASR engine to process any audio currently buffered in the VAD/silence-detection pipeline and emit a transcription result immediately, without waiting for the next speech-end event. ###### Translations - [List available translation models](https://dragonwingdocs.qualcomm.com/api-reference/translations/list-available-translation-models.md): Returns available translation models with supported language pairs and parameters. - [Translate text between languages](https://dragonwingdocs.qualcomm.com/api-reference/translations/translate-text-between-languages.md): Translates text from a source language to a target language. - [Close active translation session](https://dragonwingdocs.qualcomm.com/api-reference/translations/close-active-translation-session.md): Ends the current translation session and releases server resources. ###### TTS - [List available TTS models and voices](https://dragonwingdocs.qualcomm.com/api-reference/tts/list-available-tts-models-and-voices.md): Returns available TTS models with voices, capabilities, and supported sample rates. Filter by language using the `language` query parameter. - [Synthesize speech from text](https://dragonwingdocs.qualcomm.com/api-reference/tts/synthesize-speech-from-text.md): Converts text into spoken audio using text-to-speech synthesis. - [Close active text-to-speech session](https://dragonwingdocs.qualcomm.com/api-reference/tts/close-active-text-to-speech-session.md): Ends the current TTS session and releases server resources. - [Cancel an active text-to-speech playback](https://dragonwingdocs.qualcomm.com/api-reference/tts/cancel-an-active-text-to-speech-playback.md): Immediately stops an in-progress TTS synthesis or playback. The session remains open. ###### WebSocket - [Upgrade to WebSocket for real-time streaming](https://dragonwingdocs.qualcomm.com/api-reference/websocket/upgrade-to-websocket-for-real-time-streaming.md): Upgrades the HTTP connection to a WebSocket for real-time audio streaming. ##### Robotics workflows - [Robotics Workflow Overview](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/robotics-workflows_overview.md) - [Software Setup](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/software-setup.md) - [Depth Estimation on the NPU](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/npu-workflows.md): Build a ROS 2 depth estimation node that runs a quantized TFLite model on the Hexagon HTP NPU via the QNN delegate. ###### QRB ROS ecosystem - [QRB ROS Ecosystem](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/qrb-ros-overview.md) - [QRB ROS Samples](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/qrb-ros-samples.md) - [QRB ROS Transport](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/qrb-ros-transport.md) - [QRB ROS NN Inference](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/qrb-ros-nn-inference.md) - [QRB ROS Simulation](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/qrb-ros-simulation.md) - [QRB ROS Benchmark](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/qrb-ros-benchmark.md) - [ROS 2 DDS Config Optimizer](https://dragonwingdocs.qualcomm.com/Ubuntu/robotics-workflows/qrb-ros-dds-optimizer.md) ##### Tools - [Develop with Visual Studio Code via SSH](https://dragonwingdocs.qualcomm.com/Ubuntu/tools/vscode-ssh-setup.md) - [Docker](https://dragonwingdocs.qualcomm.com/Ubuntu/tools/docker.md) - [Profiling models](https://dragonwingdocs.qualcomm.com/Ubuntu/tools/profiling-models.md) - [libqcperf](https://dragonwingdocs.qualcomm.com/Ubuntu/tools/libqcperf.md): A lightweight C library for real-time performance monitoring on Qualcomm chipsets, with pluggable backends for CPU, NPU, and power/thermal metrics. ##### Legal - [Legal Information](https://dragonwingdocs.qualcomm.com/Ubuntu/Legal_notice.md) ## OpenAPI Specs - [openapi](/Ubuntu/ai-workflows/api-specs/llm-vlm-service/openapi.yaml) - [openapi](/Ubuntu/ai-workflows/api-specs/audio-analytics/openapi.yaml) ## Optional - [RB3 Gen 2 Dev Kit](https://docs.qualcomm.com/nav/home?product=1601111740013077) - [RB3 Gen 2 Lite Dev Kit](https://docs.qualcomm.com/nav/home?product=1601111740013077) - [IQ-X7181 EVK](https://docs.qualcomm.com/doc/80-77183-262/topic/ubuntu-qsg-landing-page-1.html)