Dual AI Studio NDI | AI coaching for live TV

Dual AI Studio NDI Edition is designed for television and web TV teams, studios, directors, technicians, presenters, trainers and creators working with an NDI video workflow on a Mac. It is particularly useful when a live production needs to monitor technical picture and sound quality while also supporting on-air performance.

The software provides continuous technical and performance guidance without pulling the team away from the live programme. It can highlight points related to framing, lighting, sound, diction, pace or engagement, then deliver those cues visually or through an earpiece. The goal is to help the presenter and production team react faster while remaining focused on the live output.

The Tech assistant focuses on technical aspects of the feed, including framing, lighting and sound. The Perf assistant focuses on the presenter's delivery, such as diction, pace and engagement. Each assistant can use its own settings, feedback window and audio output.

DUAL mode runs the Tech and Perf assistants in parallel to provide two complementary analyses during the same session. This lets you receive technical guidance and performance coaching at the same time, with separate visual and audio feedback. It is the most complete mode for continuous live-production support.

No. Dual AI Studio NDI Edition is an assistance and coaching tool that complements a production team; it does not make editorial or technical decisions on the team's behalf. AI recommendations should always be interpreted in the context of the programme, its creative choices and the constraints of the live production.

The software can be configured for different contexts, including news programmes, interviews, gaming productions and online live streams. A free-text context field also lets you describe the session more precisely so the AI assistants can adapt their guidance. The trial version is the best way to validate your real production format before purchasing.

Dual AI Studio receives NDI video and audio sources available on the local network, analyses the signal, then publishes separate feedback channels for use in the control room. NDI inputs and outputs can be enabled individually, and the receive mode can be adjusted in the interface. This makes it possible to add AI coaching to an existing workflow without rebuilding the entire production chain.

Yes. Dual AI Studio is designed to analyse NDI feeds published by production software such as OBS, vMix and Wirecast. The source must be available over NDI in the same network environment. Testing the complete chain — video, audio, feedback routing and latency — with the trial version is recommended before deployment.

You can select an NDI video source, an NDI audio source and a receive mode suited to your network. Dual AI Studio generates separate Tech, Perf and Transcription feedback outputs that can be enabled individually. Audio streams can also be routed separately to match the requirements of the production.

IFB, or Interruptible Foldback, is an audio return used in television to send cues to a presenter through an earpiece. Dual AI Studio can convert Tech and Perf recommendations into synthesised voices and route them to configurable audio outputs. Levels can be adjusted, and the two assistants can be sent to separate outputs.

Yes. Feedback windows can be configured separately with a background colour, text colour, font and font size suited to the studio or monitor. They can be used as control windows, displayed on dedicated screens or integrated into a video composition.

Yes. Feedback can be published as NDI streams, and a green background can be applied to the windows for chroma-key compositing in the production software. This makes it possible to use the information as a control-room aid, a studio display or an on-air graphic element, depending on the chosen workflow.

Yes. The interface lets you select image sources, NDI audio sources and microphone inputs, then assign the Tech and Perf voices to the required outputs. This separation makes the software adaptable to a camera, an OBS programme feed, a screen share or a studio audio configuration.

Dual AI Studio supports both approaches. Cloud mode uses Gemini with an API key, while local mode connects to an Ollama server and a model installed on your machine. You can favour the convenience and performance of the cloud or keep AI inference on your own workstation when data control is the priority.

Select Ollama as the AI source, choose a compatible local model and enter the server address — the manual uses http://localhost:11434 as the default. A connection test is available before startup. Local inference requires a sufficiently powerful machine, particularly in terms of processor, graphics resources and memory.

Select Cloud mode, choose an available Gemini model and enter your own API key. The application provides access to Google AI Studio where the key can be obtained. Before production use, check the terms, quotas and any charges attached to your AI provider account.

Yes. The API settings module lets you switch between Gemini and Ollama, change the model, update the API key or local server address, then test and save the configuration. This flexibility helps match the AI engine to the available computing power, privacy requirements and production needs.

Ollama mode lets you run AI inference on your own machine instead of sending the analysis to a cloud AI model. It is the preferred option when control over processing is essential. Overall confidentiality still depends on the complete architecture — including the NDI network, storage, audio services, activation and updates — and should be reviewed against your organisation's requirements.

Local models require a powerful machine with a capable processor, suitable graphics resources and a significant amount of RAM. Exact requirements depend on the selected Ollama model, video resolution and analysis frequency. Test the actual production configuration with the trial version before choosing the full licence.

Yes. Tech and Perf analysis frequencies can be adjusted in frames per minute, and auto-coaching can trigger analysis according to the selected cadence. This lets you balance responsiveness, machine load and use of the selected cloud model.

The Intelligence & Context section lets you describe the session in a free-text field and provides shortcuts such as News, Gaming and Interview. A precise context helps the assistants produce guidance that is more relevant to the programme format, tone and live-production objectives.

Yes. The interface is fully available in French and English, making it suitable for bilingual or international teams. Language selection is also available during activation and initial configuration.

Yes. The recording module lets you choose the resolution, PAL or NTSC video standard, frame rate, destination folder and recording name before starting. A recorded session can then support review, training, quality control or editorial reuse.

Yes. Dual AI Studio includes an RTMP streaming module where you enter the platform URL and stream key. Resolution, frame rate, video bitrate and audio bitrate can be configured before broadcasting to a compatible platform such as YouTube or Twitch.

The console displays system status, potential errors and AI-engine connection checks. It also brings together text generated by the Tech and Perf assistants, the presenter's live transcript and information about the data sent for analysis. Logs can be exported and automatic saving is enabled by default, which supports diagnostics and post-session review.

Dual AI Studio NDI Edition is exclusively compatible with Apple Silicon Macs: M1, M2, M3, M4 and later generations. Intel Macs are not supported. A minimum of 16 GB of RAM is required to run the software.

When purchasing, first launch the software to retrieve your Hardware ID, then provide this identifier in the form. After purchase, enter in the software the email address used for the purchase and the licence serial number. The full licence is a single-seat licence for one Mac.

The trial includes access to the NDI interface, DUAL Tech and Perf mode, local and cloud AI, IFB audio feedback, auto-coaching and the bilingual interface. It is designed to test the application in your real production environment before purchase. Published limitations include a visible watermark, limited usage duration, no automatic updates and limited technical support.

The full licence provides unlimited access to the features, unrestricted DUAL mode, local and cloud AI, NDI outputs, IFB feedback and auto-coaching. It removes the watermark, covers one Mac and includes one year of updates. It is intended for professionals who want to integrate AI coaching into their production workflow on an ongoing basis.

Start with the trial to confirm compatibility with your Mac, NDI network, audio outputs and chosen AI engine. Once the workflow is validated, move to the full licence to remove the limitations and receive unlimited use with one year of updates.