Add post-call delivery configuration and related models to VoiceAgent… by howieleung · Pull Request #49813 · Azure/azure-sdk-for-python

Howie Leung (howieleung) · 2026-11-03T19:58:18Z

Description

Please add an informative description that covers that changes made by the pull request and link all relevant issues.

If an SDK is being regenerated based on a new API spec, a link to the pull request containing these API spec changes should be included above.

All SDK Contribution checklist:

The pull request does not introduce [breaking changes]
CHANGELOG is updated for new features, bug fixes or other significant changes.
I have read the contribution guidelines.

General Guidelines and Best Practices

Title of the pull request is clear and informative.
There are a small number of commits, each of which have an informative message. This means that previously merged commits do not appear in the history of the PR. For more information on cleaning up the commits in your PR, see this page.

Testing Guidelines

Pull request includes test coverage for the included changes.

…#49805)

azure-pipelines · 2026-11-03T19:59:01Z

Azure Pipelines: Successfully started running 1 pipeline(s). 9 pipeline(s) were filtered out due to trigger conditions. There may be pipelines that require an authorized user to comment /azp run to run.

Copilot

🟡 Changes recommended

The new post-call delivery feature lacks a required changelog entry covering VoiceAgentDefinition.post_call, Event Grid/routine destinations, and the preview restriction on audio delivery, leaving the capability and limitation out of release notes.

1 open issue

Add changelog entry for post-call delivery feature · New

What changed in this PR

Adds post-call delivery configuration to Voice Agents in azure-ai-projects, allowing callers to configure Event Grid and routine destinations.

Changes:

Adds VoiceAgentDefinition.post_call, destination models, and related enums; documents that audio delivery is unsupported in this preview.
Exports the new types and refreshes API snapshots, cross-language mappings, and the TypeSpec revision.
Adds Pylint suppressions to the sample CI runner.

File	Description
`sdk/ai/azure-ai-projects/tsp-location.yaml`	Updates the source specification commit.
`sdk/ai/azure-ai-projects/tests/samples/ci/run_samples.py`	Suppresses line-length and unused-suppression warnings.
`sdk/ai/azure-ai-projects/azure/ai/projects/models/_models.py`	Adds post-call configuration and destination models.
`sdk/ai/azure-ai-projects/azure/ai/projects/models/_enums.py`	Adds destination and delivery-content enums.
`sdk/ai/azure-ai-projects/azure/ai/projects/models/__init__.py`	Exports the new models and enums.
`sdk/ai/azure-ai-projects/apiview-properties.json`	Updates cross-language API mappings and version.
`sdk/ai/azure-ai-projects/api.metadata.yml`	Refreshes the API hash.
`sdk/ai/azure-ai-projects/api.md`	Records the expanded public API.

_{Review effort: Balanced}

Copilot

🟡 Changes recommended

The new post_call API lacks tests for serialization and polymorphic decoding of both destination subtypes, and release notes omit the feature and its preview audio restriction.

2 open issues

_{Review effort: Balanced}

+    destinations: list["_models.VoiceAgentPostCallDestination"] = rest_field(
+        visibility=["read", "create", "update", "delete", "query"]
+    )


- Improved the clarity of SSE frame documentation in AgentsOperations by formatting the description of the `data` field. - Expanded the list of known values for the `provider` keyword in BetaVoiceAgentsTelephonyOperations to include "sip" for both bindings and calls. - Updated the commit reference in tsp-location.yaml to reflect the latest changes in the Azure REST API specs repository.

Copilot

🟡 Changes recommended

The new handoff events return dictionaries instead of typed models in both sync and async dispatch, and transfer_tool annotations incorrectly exclude valid null values.

4 open issues

_{Review effort: Balanced}

+    VoiceAgentServerEventSessionHandoffAborted,
+    VoiceAgentServerEventSessionHandoffCompleted,
+    VoiceAgentServerEventSessionHandoffStarted,


- Introduced new methods for managing live conversations in the BetaVoiceAgentsLiveConversationsOperations class, including listing, retrieving, deleting, and streaming audio for live conversations. - Updated public methods documentation to reflect the addition of 9 new methods under the `beta.voice_agents.live_conversations` namespace. - Updated the TSP location YAML file to point to the latest commit in the Azure REST API specs repository.

Copilot

🟡 Changes recommended

The new realtime events are missing from both sync and async client dispatch maps and event unions, causing recv() to return dictionaries instead of models and breaking typed event handling.

12 open issues

1 file not reviewed

eng/emitter-package-lock.json: Generated file

_{Review effort: Balanced}

…renceGraph script

Copilot

🔵 Needs a closer look

The new campaigns operations class is missing from the public imports and all, causing its documented public import to fail.

11 open issues

1 resolved since last review

Document release features and preview-model breaking changes

1 file not reviewed

eng/emitter-package-lock.json: Generated file

_{Review effort: Balanced}

* [azure-ai-projects] Add GPT Live WebSocket clients and samples Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix GPT Live event registry type narrowing Validate generated event registry values through a Mapping of objects so Pyright correctly narrows model classes without casts. Add invalid-registry regression coverage. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * [azure-ai-projects] Add GPT Live REST conversation samples Add sync and async REST-only samples for stored conversation, item, and recording APIs, with opt-in downloads and deletion. Add 42 enabled offline guards and related documentation. Verified all 18 sync/async REST targets against the real service and 555 focused unit cases. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * [azure-ai-projects] Address GPT Live review feedback Protect input WAV recordings from hard-linked output aliases, normalize mapping inputs in Live serialization, and await cancellation-safe shared asynchronous transport cleanup. Add deterministic regressions for all three findings while preserving existing close-error behavior. Verified 580 focused unit cases and type/lint/build checks. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * [azure-ai-projects] Add microphone mode to GPT Live audio sample Extend the existing audio sample with optional PyAudio microphone capture and speaker playback, preserving WAV mode. Bound device buffering, surface callback/send errors, report actual audio bytes, and gracefully close the session on Ctrl+C. Add 24 microphone guards and related documentation. Verified 162 focused tests, both CLI modes against the live service, and user-confirmed physical microphone operation. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Add recorded GPT Live REST coverage Cover all nine dedicated Live REST methods in sync and async Test Proxy lifecycles, including pagination, audio content, HTTP statuses, deletion and safe fixture cleanup. Reference the published recording assets and add offline playback regressions. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Align GPT Live samples with review conventions Consolidate Live samples, remove CLI arguments, use user-temporary audio output, and cover resource ownership and sample conventions. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Make GPT Live conversation REST sample self-contained Create and clean up a stored greeting conversation when no existing ID is supplied. Preserve read-only behavior for supplied IDs and remove source-shape and filename assertions per review while retaining behavioral coverage. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Simplify GPT Live REST sample test coverage Reuse the existing REST fixture for success and failure cleanup checks. Remove the simulated greeting, clock, and persistence test setup without changing the sample. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Simplify handwritten VoiceAgents documentation Clarify Realtime and GPT Live setup, sample choices, and client descriptions without changing runtime behavior, public APIs, generated code, or tests. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * [azure-ai-projects] Align Live WebSocket support with regenerated SDK Use canonical Realtime/Live query parameters, native discriminator readback, and persisted Live delegation models. Update sync/async documentation, sample constructors and contract tests, and classify never-released Live changes as new features in the changelog. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * [azure-ai-projects] Align custom Live clients and recording layout with vnext Use native session-update delegation models, preserve typed Foundry ownership and recorded Live REST coverage, expose Live conversation operations, and add typed Realtime handoff dispatch. Refresh API artifacts and unreleased sample notes. Rebase and revalidate this local checkpoint against the newly advanced vnext target before publication. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * [azure-ai-projects] Complete vnext event alignment and refresh API Expose typed system-tool argument and input-collection events in both Realtime clients, preserving native vnext model names and generated output. Verify the rebased custom SDK with 1,973 passing tests and 120 existing skips, refresh the exact-wheel paired API artifacts, and keep never-released Live changes under features. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix Voice sample docs, streaming tests, and spelling scope Use absolute source links for strict Sphinx builds, make sync and async connection-error tests deterministic behind CI proxies, and keep voice spelling additions in the AI service configuration. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Expand VoiceAgents samples and regression coverage Add async WebSocket and sync REST examples, native feature and recorded CRUD tests, and protocol regression coverage. Fix the Live connection sample audio clock, simplify sample tests and documentation, and reference verified sanitized recordings. The generated-audio artifact test remains enabled; its recording is pending because the live service did not publish the artifact during validation. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Deduplicate VoiceAgents samples by protocol Keep one example per scenario: synchronous REST and asynchronous WebSocket workflows. Remove duplicate counterparts, convert the remaining WebSocket entrypoints, and migrate workflow regressions and documentation to the retained examples. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Simplify Voice Agent samples and fix CI validation Bundle a synthetic spoken-question WAV with shared optional input-path handling, simplify the Live function-tool workflow and documentation, and remove source-shape sample tests while retaining behavioral coverage. Align the Projects development version and API metadata with 2.11.0 Unreleased, and isolate mocked management-policy requests from inherited CI proxy settings. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Keep Voice Agent PR changes scoped to AI Remove the out-of-scope Azure Core streaming and management-policy test changes. Restore all three Core test files to the PR target while preserving every AI change and the latest target-branch merge. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Fix interrupted generated-audio recording test Wait for the persisted interrupted conversation item without requiring an optional generated-audio artifact. Validate the documented not-found result after incomplete generation, while retaining full metadata and WAV validation when audio is returned. Reference the genuine sanitized cancellation recording and preserve deletion and error-propagation checks. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --------- Co-authored-by: Xiting Zhang <xitzhang@microsoft.com> Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot

🟡 Changes recommended

The samples do not catch asyncio.TimeoutError on supported Python 3.10, causing normal receive timeouts to abort the live connection or skip timeout messaging and response cancellation.

12 open issues

1 file not reviewed

eng/emitter-package-lock.json: Generated file

_{Review effort: Balanced}

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

Copilot

🔵 Needs a closer look

Human assessment is needed because Python 3.10 timeout handling and response-limit boundary behavior introduce compatibility and unnecessary backend-work concerns.

10 open issues

2 resolved since last review

1 file not reviewed

eng/emitter-package-lock.json: Generated file

_{Review effort: Balanced}

- Introduced TelephonyCampaignDatasetReference model to represent dataset references in telephony campaigns. - Updated DatasetReference model to require a concrete version. - Refactored TelephonyCampaign to use TelephonyCampaignDatasetReference for recipient_dataset. - Implemented BetaVoiceAgentsCampaignsOperations for managing telephony campaigns, including create, cancel, and delete operations. - Added tests for campaign long-running operations (LRO) to ensure correct behavior and error handling. - Updated public methods documentation to include new campaign operations. - Cleaned up sample scripts and test files for consistency and readability.

Copilot

🟡 Changes recommended

Timeout handling remains incomplete: a stalled audio send can bypass the overall deadline, while Python 3.10 receive timeouts bypass cancellation and terminate the sample with a traceback.

10 open issues

1 resolved since last review

Add changelog entry for post-call delivery feature

1 file not reviewed

eng/emitter-package-lock.json: Generated file

_{Review effort: Balanced}

+    while True:
+        remaining_time(deadline, event_type)
+        if time.monotonic() >= next_frame:
+            await connection.session.input_audio.append(audio=silence)


github-actions · 2026-11-06T19:51:41Z

[Pilot] PR Pipeline Failure Analysis

What failed

Pipeline build 7048526 (python - pullrequest) failed across multiple jobs:

Build Analyze — a spell-check (cspell) failure on a file added/modified by this PR:
sdk/ai/azure-ai-projects/tests/agents/test_voice_agent_campaigns_lro.py:349:50 - Unknown word (Acancel).
This is a genuine unrecognized token (Acancel, likely a typo or missing word-boundary, e.g. cancel glued to a prefix) introduced by the PR's new voice-agent test file, causing the Analyze job to fail and exit PowerShell with code 1.
Build Test ubuntu2404_311_ib1 and Build Test windows2022_312_ib1 (both whl and sdist variants) — azure-core test failures unrelated to this PR's diff:
- tests/test_rest_stream_responses.py::test_iter_with_error
- tests/async_tests/test_rest_stream_responses_async.py::test_iter_with_error
Both raise ValueError: Should error before entering from a raise ValueError(...) placed inside the test body itself (not library code), at test_rest_stream_responses.py:64 and test_rest_stream_responses_async.py:56, under a with pytest.raises(ServiceRequestError): block for a request to (doesnotexist/redacted) This indicates the azure-coreHTTP client did not raiseServiceRequestErroras expected for an unresolvable host, so execution fell through to the test's own deliberateraise ValueError. This failure is identical across both OS platforms and both package formats (whl/sdist), consistent with a shared root cause in azure-core's request-error handling rather than anything touched by this PR (which only changes sdk/ai/azure-ai-projects`).
Build Test Ubuntu2404_314 also reported FAILURE on the PR checks list, consistent with the same cascading Build Analyze cspell failure blocking the stage.

Relevant pipeline output

2026-11-06T19:34:04.4747940Z ##[error]sdk/ai/azure-ai-projects/tests/agents/test_voice_agent_campaigns_lro.py:349:50 - Unknown word (Acancel)
2026-11-06T19:34:04.4750546Z ##[error]Spelling errors detected. To correct false positives or learn about spell checking see: https://aka.ms/azsdk/engsys/spellcheck
2026-11-06T19:34:04.5601616Z ##[error]PowerShell exited with code 1.

tests/test_rest_stream_responses.py:64: ValueError
E   ValueError: Should error before entering

Recommended next steps

Fixable (Build Analyze / cspell): Correct or add the unrecognized word Acancel at sdk/ai/azure-ai-projects/tests/agents/test_voice_agent_campaigns_lro.py:349:50 — likely a typo missing a space/underscore (e.g. Acancel → A cancel or a_cancel), or add it to the repo's cspell dictionary if it is an intentional identifier.
Not fixable by this PR's code change (azure-core test failures): These failures occur in sdk/core/azure-core tests, a package this PR does not modify, and reproduce identically on both OS platforms and distribution formats — a strong signal of a pre-existing/flaky azure-core issue unrelated to this PR. Recommend retrying the pipeline and, if it persists, filing/triaging against azure-core's ServiceRequestError handling for unreachable hosts separately from this PR.
See https://aka.ms/ci-fix

Automated fix: Requested

Generated by Pipeline Analysis Next Steps · copilot · sonnet50 · 52.6 AIC · ⌖ 0.573 AIC · ⊞ 9.1K · ◷

…ust release date, and enhance sample updates

Add post-call delivery configuration and related models to VoiceAgent (…

477bf14

…#49805)

Copilot AI balanced review requested due to automatic review settings November 3, 2026 19:58

Howie Leung (howieleung) requested review from a team, Darren Cohen (dargilco), Glenn Harper (glharper), kingernupur, Nikolay Rovinskiy (nick863) and Travis Angevine (trangevi) as code owners November 3, 2026 19:58

github-actions Bot added the AI Projects label Nov 3, 2026

Copilot started reviewing on behalf of Howie Leung (howieleung) November 3, 2026 19:59

View session

Copilot AI reviewed Nov 3, 2026

View reviewed changes

Comment thread sdk/ai/azure-ai-projects/azure/ai/projects/models/_models.py Resolved

This comment has been minimized.

Sign in to view

Howie Leung (howieleung) closed this Nov 3, 2026

Howie Leung (howieleung) reopened this Nov 4, 2026

Howie Leung (howieleung) added 2 commits November 3, 2026 16:14

Merge branch 'main' into feature/azure-ai-projects/vnext

5617188

Merge branch 'main' into feature/azure-ai-projects/vnext

b97869f

Copilot AI balanced review requested due to automatic review settings November 4, 2026 19:22

Copilot started reviewing on behalf of Howie Leung (howieleung) November 4, 2026 19:22

View session

Copilot AI reviewed Nov 4, 2026

View reviewed changes

Comment thread sdk/ai/azure-ai-projects/azure/ai/projects/models/_models.py

Comment on lines +29806 to +29808

destinations: list["_models.VoiceAgentPostCallDestination"] = rest_field(

visibility=["read", "create", "update", "delete", "query"]

)

Copilot AI balanced review requested due to automatic review settings November 4, 2026 20:00

Copilot started reviewing on behalf of Howie Leung (howieleung) November 4, 2026 20:00

View session

Copilot AI reviewed Nov 4, 2026

View reviewed changes

Copilot AI balanced review requested due to automatic review settings November 4, 2026 22:08

Howie Leung (howieleung) requested review from Chenxi Jiang (ChenxiJiang333), iscai-msft, Laia Patiño (laiapat) and Yuchao Yan (msyyc) as code owners November 4, 2026 22:08

This comment has been minimized.

Sign in to view

Bump version to 2.11.0 and update telephony call job paths in tests

32aa3d5

Copilot AI balanced review requested due to automatic review settings November 5, 2026 22:57

Copilot started reviewing on behalf of Howie Leung (howieleung) November 5, 2026 22:58

View session

Copilot AI reviewed Nov 5, 2026

View reviewed changes

Comment thread sdk/ai/azure-ai-projects/CHANGELOG.md Resolved

This comment has been minimized.

Sign in to view

Update cspell configuration and timestamp format in GenerateModelRefe…

bf14fa4

…renceGraph script

Copilot AI balanced review requested due to automatic review settings November 6, 2026 00:17

Copilot started reviewing on behalf of Howie Leung (howieleung) November 6, 2026 00:17

View session

Copilot AI reviewed Nov 6, 2026

View reviewed changes

xitzhang (xitzhang) requested review from Hanchi Wang (luigiw), Neehar Duvvuri (needuv) and Ankit Singhal (singankit) as code owners November 6, 2026 01:26

Copilot AI balanced review requested due to automatic review settings November 6, 2026 01:26

Copilot started reviewing on behalf of Howie Leung (howieleung) November 6, 2026 01:27

View session

Copilot AI reviewed Nov 6, 2026

View reviewed changes

Comment thread sdk/ai/azure-ai-projects/samples/agents/voice/live_sample_utils.py Resolved

This comment has been minimized.

Sign in to view

github-advanced-security Bot found potential problems Nov 6, 2026

View reviewed changes

Comment thread sdk/ai/azure-ai-projects/samples/agents/voice/sample_voice_agent_live_read_conversation.py Dismissed

Handle both asyncio and standard TimeoutError

a68b1d5

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

Copilot AI balanced review requested due to automatic review settings November 6, 2026 02:55

Copilot started reviewing on behalf of Howie Leung (howieleung) November 6, 2026 02:55

View session

Copilot AI reviewed Nov 6, 2026

View reviewed changes

This comment has been minimized.

Sign in to view

Copilot AI balanced review requested due to automatic review settings November 6, 2026 18:58

Copilot started reviewing on behalf of Howie Leung (howieleung) November 6, 2026 18:59

View session

Copilot AI reviewed Nov 6, 2026

View reviewed changes

Comment thread sdk/ai/azure-ai-projects/samples/agents/voice/live_sample_utils.py

while True:

remaining_time(deadline, event_type)

if time.monotonic() >= next_frame:

await connection.session.input_audio.append(audio=silence)

Update CHANGELOG for version 2.11.0: refine feature descriptions, adj…

5649a95

…ust release date, and enhance sample updates

Conversation

Howie Leung (howieleung) commented Nov 3, 2026

Description

All SDK Contribution checklist:

General Guidelines and Best Practices

Testing Guidelines

Uh oh!

azure-pipelines Bot commented Nov 3, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

Uh oh!

Uh oh!

This comment has been minimized.

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

Uh oh!

Uh oh!

This comment has been minimized.

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

Uh oh!

Uh oh!

This comment has been minimized.

Copilot AI left a comment

Choose a reason for hiding this comment

🔵 Needs a closer look

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

Uh oh!

Uh oh!

This comment has been minimized.

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🔵 Needs a closer look

Uh oh!

This comment has been minimized.

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

Uh oh!

github-actions Bot commented Nov 6, 2026

What failed

Recommended next steps

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants