
Digital health clinical asr eval
Scores clinical ASR transcripts with keyword-focused error metrics and a leaderboard
- Scripts
- None
- Platforms
- Claude Code, Codex, Cursor
- GitHub stars
- 4K
- Last commit
- 2 days ago
- License
- Apache-2.0
- Listing updated
- 1 day ago
Similar skills
About Digital health clinical asr eval
Digital health clinical asr eval is an open-source agent skill for health and science. It is available on Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf and Goose.
This skill takes a NeMo-format manifest of synthetic clinical audio clips, sends each clip to a hosted speech recognition model, and compares the transcripts against the reference text. It computes word, character, keyword and sentence error rates, with keyword error rate treated as the main signal because it checks whether flagged medical terms come out as contiguous, correctly spelled words.
The results are arranged into a five-section markdown leaderboard that breaks scores down by entity category, pronunciation source, noise level and individual term. A decision guide then suggests whether to expand the term list, fine-tune the model, or harden the evaluation, and it flags pronunciation-coverage gaps as a separate problem from model accuracy.
Users need an NVIDIA API key, the NeMo Riva client and soundfile installed, and audio files present on disk. The workflow discloses that each clip and its reference text are sent to an external NVIDIA service, so it should only be used with synthetic audio rather than real patient recordings. Scoring itself runs locally in pure Python.
When your agent uses it
Stage 3 of Clinical ASR Flywheel. Score a NeMo manifest, produce the five-section KER leaderboard (by-ipa_source diagnostic). Not for ASR auth (/riva-asr).
From the skill's SKILL.md. Your agent reads this description to decide when to load the skill.
What it can access
From a static check of its 7 files (48 KB) on GitHub on 17 November 2026. Nothing was run. Read the files before you install a skill: it can tell your agent to run anything its tools allow.
Check another skill before you install it.
Install
Copy the skill folder to ~/.claude/skills/digital-health-clinical-asr-eval to use it in all your projects:
tmp=$(mktemp -d) &&
git clone --depth 1 https://github.com/NVIDIA/skills.git "$tmp" &&
mkdir -p ~/.claude/skills &&
rm -rf ~/.claude/skills/digital-health-clinical-asr-eval &&
cp -R "$tmp/skills/digital-health-clinical-asr-eval" ~/.claude/skills/digital-health-clinical-asr-eval &&
rm -rf "$tmp"For one project only, put the folder in .claude/skills/digital-health-clinical-asr-eval inside it instead.
Copy the skill folder to ~/.agents/skills/digital-health-clinical-asr-eval to use it in every repository:
tmp=$(mktemp -d) &&
git clone --depth 1 https://github.com/NVIDIA/skills.git "$tmp" &&
mkdir -p ~/.agents/skills &&
rm -rf ~/.agents/skills/digital-health-clinical-asr-eval &&
cp -R "$tmp/skills/digital-health-clinical-asr-eval" ~/.agents/skills/digital-health-clinical-asr-eval &&
rm -rf "$tmp"For one repository only, put the folder in .agents/skills/digital-health-clinical-asr-eval inside it instead.
Copy the skill folder to ~/.cursor/skills/digital-health-clinical-asr-eval to use it in every project:
tmp=$(mktemp -d) &&
git clone --depth 1 https://github.com/NVIDIA/skills.git "$tmp" &&
mkdir -p ~/.cursor/skills &&
rm -rf ~/.cursor/skills/digital-health-clinical-asr-eval &&
cp -R "$tmp/skills/digital-health-clinical-asr-eval" ~/.cursor/skills/digital-health-clinical-asr-eval &&
rm -rf "$tmp"For one project only, put the folder in .cursor/skills/digital-health-clinical-asr-eval inside it instead.
Run in a terminal (add --scope workspace to install it for the current workspace only):
gemini skills install https://github.com/NVIDIA/skills.git --path skills/digital-health-clinical-asr-evalOr copy the folder to ~/.gemini/skills/digital-health-clinical-asr-eval yourself.
For GitHub Copilot in VS Code, copy the skill folder to ~/.copilot/skills/digital-health-clinical-asr-eval:
tmp=$(mktemp -d) &&
git clone --depth 1 https://github.com/NVIDIA/skills.git "$tmp" &&
mkdir -p ~/.copilot/skills &&
rm -rf ~/.copilot/skills/digital-health-clinical-asr-eval &&
cp -R "$tmp/skills/digital-health-clinical-asr-eval" ~/.copilot/skills/digital-health-clinical-asr-eval &&
rm -rf "$tmp"For one project only, put the folder in .github/skills/digital-health-clinical-asr-eval inside it instead.
Get the folder:
git clone --depth 1 https://github.com/NVIDIA/skills.gitThen copy skills/skills/digital-health-clinical-asr-eval into your agent's skills folder as digital-health-clinical-asr-eval. Agents that support SKILL.md read one folder per skill.
What it does
- Keyword error rate
- Checks whether flagged clinical terms appear intact and in order in each transcript.
- Standard error metrics
- Reports word, character and sentence error rates alongside the keyword metric.
- Five-section leaderboard
- Produces a fixed-order markdown report covering headline scores and breakdowns.
- Pronunciation source split
- Compares results by pronunciation source to reveal coverage gaps in the pipeline.
- Post-eval routing
- Suggests whether to expand the term list, fine-tune the model, or stop and harden the evaluation.
- Hosted ASR transcription
- Transcribes each manifest row through a configurable NVIDIA hosted or self-hosted speech model.
Good for
- Score a manifest of synthetic clinical audio against reference transcripts
- Find which drug or procedure terms a speech model misrecognizes
- Compare keyword error rates across noise levels for clinical dictation
- Decide whether a speech model needs fine-tuning or more term coverage
- Generate a markdown leaderboard for a clinical ASR evaluation cycle
Digital health clinical asr eval FAQ
What does the skill need before it runs?
A NeMo-format manifest with clinical fields, an NVIDIA_API_KEY, the nvidia-riva-client and soundfile packages, and the audio files present on disk.
Does it send audio anywhere?
Yes. Each clip and its reference text are sent to an NVIDIA NVCF speech service for transcription. The skill recommends using only synthetic audio, not real patient data.
Does it generate audio or manifests?
No. If the manifest is missing or empty, the skill points the user to a separate build skill for creating one.
Which speech model does it use by default?
The default is the hosted Parakeet TDT 0.6B v2 model, and environment variables can switch to another hosted or self-hosted model.
How do I install Digital health clinical asr eval in Claude Code?
Copy the skill folder from GitHub to ~/.claude/skills/digital-health-clinical-asr-eval to use it in all your projects, or to .claude/skills/digital-health-clinical-asr-eval in one project.
Does Digital health clinical asr eval run scripts?
No. It is instructions only, with no scripts. This comes from a static check of the files on GitHub; nothing was run. Read the files before you install it.
Which AI agents can use Digital health clinical asr eval?
Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf and Goose. Each reads skills in the SKILL.md format from its own folder; the install steps on this page list them.
What are the best alternatives to Digital health clinical asr eval?
The closest alternatives to Digital health clinical asr eval are Digital health clinical asr build, Digital health clinical asr setup, Digital health clinical asr finetune, Ambient healthcare agent with nemotron voice agent and Codonfm score.
Where this comes from
- Description: from the skill's own SKILL.md.
More from NVIDIA/skills

Guides an agent in converting cuTile GPU kernels to Triton and debugging the result
- Open source
- 4K GitHub stars
- Developer tools

Adds autotuning with cached config selection to CuTile GPU kernels
- Open source
- 4K GitHub stars
- Developer tools

Guides an agent to write, debug and validate cuTile GPU kernels in Python
- Open source
- 4K GitHub stars
- Developer tools

Guides an agent through iterative profiling and tuning of cuTile GPU kernels in TileGym
- Open source
- 4K GitHub stars
- Developer tools

Guides an agent to swap TileGym cuTile kernels into Hugging Face transformers models
- Open source
- 4K GitHub stars
- Developer tools

Composes and deploys an NVIDIA VSS vision stack with captioning, detection, search
- Open source
- 4K GitHub stars
- Design and media
