Add AM62D Audio AI documentation (OS=edgeai) by junechul · Pull Request #813 · TexasInstruments/processor-sdk-doc

junechul · 2026-10-12T18:59:17Z

Summary

Adds a new Audio AI documentation set for AM62D (AM62DX). It builds with DEVFAMILY=AM62DX OS=edgeai.

What's included

New pages under source/devices/AM62DX/edgeai/:

Introduction & Overview: Audio AI on AM62D with the C7x NPU, plus a TVM + TIDL flow diagram. The workflow is framed as bring-your-own-model (BYOM).
Audio AI Model Zoo: on-target demos and the model zoo link.
Model Compile / Evaluate: compile and evaluate models with the TVM + TIDL flow.
Model Inspect: inspect compiled models.
Performance: benchmark tables (To be updated).

Supporting changes:

configs/AM62DX/AM62DX_edgeai_{config,tags,toc} for the new edgeai OS target.
A conf.py update.
Vale vocabulary additions in .github/styles/config/vocabularies/PSDK/accept.txt.
Diagram sources and exports in source/images/audioai/ (drawio, svg, png, export script).

Testing

Built locally with:
docker run -it --rm -v "$PWD":/build ghcr.io/texasinstruments/processor-sdk-doc:latest make DEVFAMILY=AM62DX OS=edgeai

Notes

Only AM62DX/edgeai files and shared config entries are touched. Other devices are not affected.

github-actions · 2026-10-12T19:00:06Z

New warnings found with rstcheck:

source/devices/AM62DX/edgeai/performance.rst:68: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_VGGISH11__".
source/devices/AM62DX/edgeai/performance.rst:74: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_YAMNET__".
source/devices/AM62DX/edgeai/performance.rst:80: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_GCRN__".
source/devices/AM62DX/edgeai/performance.rst:86: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_GTCRN__".
source/devices/AM62DX/edgeai/model_compile_evaluate.rst:67: (ERROR/3) Undefined substitution referenced: "__TIDLRUNNER_REPO_URL__".
source/devices/AM62DX/edgeai/model_compile_evaluate.rst:179: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_VGGISH11__".
source/devices/AM62DX/edgeai/model_compile_evaluate.rst:184: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_YAMNET__".
source/devices/AM62DX/edgeai/model_compile_evaluate.rst:189: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_GCRN__".
source/devices/AM62DX/edgeai/model_compile_evaluate.rst:194: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_GTCRN__".
source/devices/AM62DX/edgeai/model_zoo.rst:38: (ERROR/3) Undefined substitution referenced: "__DATASET_AUDIO_CLS__".
source/devices/AM62DX/edgeai/model_zoo.rst:39: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_VGGISH11__".
source/devices/AM62DX/edgeai/model_zoo.rst:43: (ERROR/3) Undefined substitution referenced: "__DATASET_AUDIO_CLS__".
source/devices/AM62DX/edgeai/model_zoo.rst:44: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_YAMNET__".
source/devices/AM62DX/edgeai/model_zoo.rst:48: (ERROR/3) Undefined substitution referenced: "__DATASET_SPEECH_ENH__".
source/devices/AM62DX/edgeai/model_zoo.rst:49: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_GCRN__".
source/devices/AM62DX/edgeai/model_zoo.rst:53: (ERROR/3) Undefined substitution referenced: "__DATASET_SPEECH_ENH__".
source/devices/AM62DX/edgeai/model_zoo.rst:54: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_GTCRN__".
source/devices/AM62DX/edgeai/model_zoo.rst:72: (ERROR/3) Undefined substitution referenced: "__MODELZOO_REPO_URL__".
source/devices/AM62DX/edgeai/model_zoo.rst:93: (ERROR/3) Undefined substitution referenced: "__TARGET_VENV_PATH__".
source/devices/AM62DX/edgeai/model_zoo.rst:164: (ERROR/3) Undefined substitution referenced: "__DATASET_AUDIO_CLS__".
source/devices/AM62DX/edgeai/model_zoo.rst:295: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_VGGISH11__".
source/devices/AM62DX/edgeai/model_zoo.rst:296: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_YAMNET__".
source/devices/AM62DX/edgeai/model_zoo.rst:297: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_GCRN__".
source/devices/AM62DX/edgeai/model_zoo.rst:304: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_GTCRN__".
source/devices/AM62DX/edgeai/model_zoo.rst:307: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_GTCRN__".
source/devices/AM62DX/edgeai/introduction.rst:9: (ERROR/3) Undefined substitution referenced: "__AUDIOAI_SDK_VERSION__".
source/devices/AM62DX/edgeai/introduction.rst:65: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_VGGISH11__".
source/devices/AM62DX/edgeai/introduction.rst:70: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_YAMNET__".
source/devices/AM62DX/edgeai/introduction.rst:75: (ERROR/3) Undefined substitution referenced: "__DATASET_AUDIO_CLS__".
source/devices/AM62DX/edgeai/introduction.rst:80: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_GCRN__".
source/devices/AM62DX/edgeai/introduction.rst:86: (ERROR/3) Undefined substitution referenced: "__MODEL_ID_GTCRN__".
source/devices/AM62DX/edgeai/introduction.rst:93: (ERROR/3) Undefined substitution referenced: "__DATASET_SPEECH_ENH__".

github-actions · 2026-10-12T19:00:31Z

New unreachable files found with check_files.py:

WARNING:root:File not used: source/images/audioai/export-diagrams.sh
WARNING:root:File not used: source/images/audioai/tvm_tidl_flow.png
WARNING:root:File not used: source/images/audioai/tvm_tidl_flow.drawio

StaticRocket

This feels like it will be blocked by #740 as well, though the user did not tag the commits.

Additionally, review https://github.com/TexasInstruments/processor-sdk-doc?tab=contributing-ov-file#commit-types

StaticRocket · 2026-10-13T20:46:04Z

Acronyms are not allowed here

Removed. accept.txt now only adds [Ss]ubgraphs?, logits?, perfsim and venv, which are the only entries Vale flagged. No acronyms.

StaticRocket · 2026-10-13T20:47:22Z

+*********************************************
+AM62D Audio AI Software Developer's Guide
+*********************************************
+


https://github.com/TexasInstruments/processor-sdk-doc?tab=contributing-ov-file#headings--sections

Done. The index title now uses #### (Section 1), and all pages follow the heading hierarchy in the contributing guide, with each adornment matching its title length.

StaticRocket · 2026-10-13T20:48:10Z

+Introduction & Overview
+#######################
+
+.. |tm| unicode:: U+2122


Our IT department is stupid so you can't access this link internally, but:

https://docutils.sourceforge.io/docs/ref/rst/definitions.html

Thanks. Replaced the custom .. |tm| unicode:: U+2122 with .. include:: <isonum.txt> and |trade| in all four edgeai pages.

StaticRocket · 2026-10-13T20:49:18Z

+********************
+Reference models
+********************
+


Title overhang

Fixed. Every over/underline in the edgeai pages now has exactly the same length as its title.

StaticRocket · 2026-10-13T20:49:57Z

+
+The end-to-end path is the same for a reference model or your own model:
+
+#. **Obtain an ONNX model** — a reference graph shipped with the AudioAI ModelZoo,


Unnecessary unicode. See #726

Replaced the em dashes and the division sign with ASCII (- and /). The edgeai .rst files are now ASCII-only.

StaticRocket · 2026-10-13T20:51:38Z

+into a dedicated Python 3.10 environment. The TIDL tools target Python 3.10.
+These steps assume `pyenv <https://github.com/pyenv/pyenv>`__.
+
+**1. Clone the repository** (|__TIDLRUNNER_REPO_URL__|):


Use autonumbering

Converted the hand-numbered steps to #. autonumbered lists, here and in model_zoo.rst.

StaticRocket · 2026-10-13T20:51:58Z

+.. code-block:: console
+
+   $ git clone https://github.com/TexasInstruments/edgeai-tidlrunner.git
+   $ cd edgeai-tidlrunner


Indent content to fall under respective list item

Done. Each step's paragraphs, code blocks and table are now indented 3 spaces under their list item.

StaticRocket · 2026-10-13T20:55:16Z

 OS = os.environ.get("OS", "")
+
+# Audio AI docs prefix every shell command with a "$ " prompt symbol; strip it
+# from the copy-button output so copied text is directly runnable. 


Just use a global regex definition: https://sphinx-copybutton.readthedocs.io/en/latest/use.html#using-regexp-prompt-identifiers

I kept this scoped to the AM62DX edgeai build on purpose. A global copybutton_prompt_text would change copy behavior for every device doc.

StaticRocket · 2026-10-13T20:56:00Z

Nope, we are not pulling this

Removed from the PR. The file is no longer tracked.

StaticRocket · 2026-10-13T20:58:00Z

Use an svg or a graphviz graph definition. We are not pulling in an electron app for building documentation.

Removed from the PR. The file is no longer tracked.

StaticRocket

This feels like it will be blocked by #740 as well, though the user did not tag the commits.

Additionally, review https://github.com/TexasInstruments/processor-sdk-doc?tab=contributing-ov-file#commit-types

Signed-off-by: JuneChul Roh <jroh@ti.com>

…M62D setup Signed-off-by: JuneChul Roh <jroh@ti.com>

Audio AI docs prefix every shell command with "$ "; configure copybutton to strip it so copied text is directly runnable. Signed-off-by: JuneChul Roh <jroh@ti.com>

Signed-off-by: JuneChul Roh <jroh@ti.com>

…tion. Signed-off-by: JuneChul Roh <jroh@ti.com>

…l zoo URL Rename source/devices/AM62DX/audioai to edgeai and the AM62DX_audioai_* config files to AM62DX_edgeai_*. Point __MODELZOO_REPO_URL__ and the model zoo clone command to TexasInstruments/audioai-modelzoo. Signed-off-by: JuneChul Roh <jroh@ti.com>

Split into "Model Compile and Evaluate" (model_compile_evaluate.rst) and "Model Inspect" (model_inspect.rst). Update the toctree, the AM62DX edgeai toc config, labels, and cross-references in the other pages. Signed-off-by: JuneChul Roh <jroh@ti.com>

Signed-off-by: JuneChul Roh <jroh@ti.com>

Trim the Vale vocabulary to the entries Vale needs, fix heading adornments and non-ASCII text, use isonum.txt and #. autonumbered steps, and replace the removed replacement variables with literal values. Signed-off-by: JuneChul Roh <jroh@ti.com>

junechul · 2026-10-14T00:20:26Z

@StaticRocket I have addressed most of your review comments. Please let me know if any further fix is required. Thanks!

junechul requested review from StaticRocket, VeeruPrudhvi, cshilwant, gehariprasath, jeevantelukula, praneethbajjuri and uditkumarti as code owners October 12, 2026 18:59

StaticRocket requested changes Oct 13, 2026

View reviewed changes

junechul added 16 commits October 13, 2026 19:07

feat(am62d-audioai): add config files and buildable doc skeleton

bf06c93

Signed-off-by: JuneChul Roh <jroh@ti.com>

feat(am62d-audioai): add Section 1 introduction with TVM+TIDL diagram

f1bd267

Signed-off-by: JuneChul Roh <jroh@ti.com>

feat(am62d-audioai): add Section 2 Audio AI Model Zoo on-target demos

4c1c7d1

Signed-off-by: JuneChul Roh <jroh@ti.com>

feat(am62d-audioai): add Section 3 compile/evaluate/inspect workflow

6f16447

Signed-off-by: JuneChul Roh <jroh@ti.com>

feat(am62d-audioai): add Section 4 performance benchmark tables

6cf0df0

Signed-off-by: JuneChul Roh <jroh@ti.com>

feat(am62d-audioai): integrate navigation, cross-links, and full build

0db7cfb

Signed-off-by: JuneChul Roh <jroh@ti.com>

docs(am62d-audioai): reframe as BYOM workflow and fix C7 NPU terminology

283617c

Signed-off-by: JuneChul Roh <jroh@ti.com>

docs(am62d-audioai): reconcile compile/eval/inspect with standalone A…

c0b6471

…M62D setup Signed-off-by: JuneChul Roh <jroh@ti.com>

docs(am62d-audioai): strip prompt symbol from copy button output

319271f

Audio AI docs prefix every shell command with "$ "; configure copybutton to strip it so copied text is directly runnable. Signed-off-by: JuneChul Roh <jroh@ti.com>

docs(am62d-audioai): update Introduction and Model Zoo sections

db96350

Signed-off-by: JuneChul Roh <jroh@ti.com>

docs(am62d-audioai): Updated "Model Compile / Evaluate / Inspect" sec…

630cb1c

…tion. Signed-off-by: JuneChul Roh <jroh@ti.com>

docs(am62d-audioai): fix inspector path and Jupyter launcher description

95ea7ac

Signed-off-by: JuneChul Roh <jroh@ti.com>

chore(am62d-audioai): stop tracking diagram sources and export script

880ba36

Signed-off-by: JuneChul Roh <jroh@ti.com>

junechul force-pushed the am62d-audioai branch from 9ac0c1e to 0f63803 Compare October 14, 2026 00:15


		The end-to-end path is the same for a reference model or your own model:

		#. Obtain an ONNX model — a reference graph shipped with the AudioAI ModelZoo,

Conversation

junechul commented Oct 12, 2026

Summary

What's included

Testing

Notes

Uh oh!

github-actions Bot commented Oct 12, 2026

Uh oh!

github-actions Bot commented Oct 12, 2026

Uh oh!

StaticRocket left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

StaticRocket left a comment

Choose a reason for hiding this comment

Uh oh!

junechul commented Oct 14, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants