Generated banner for Tts Voiceover
Generated logo for Tts Voiceover

Tts Voiceover

MicrosoftOfficialPublished by Microsoft

  • Skill
  • Tool integration
  • Automation and integrations
  • MIT
  • Low risk
  • Has scripts

From the author

Text-to-speech voice-over generation from YAML speaker notes using Azure AI Speech neural voices with SSML pronunciation control
— the description by Microsoft

AutoGineer summary

Written by AutoGineer from the skill’s own files, not by the author

This skill automates the creation of synthetic voice-over audio for slide presentations. It processes speaker notes defined in YAML files, applies custom acronym pronunciations via an SSML-based lexicon, and utilizes Azure AI Speech neural voices to generate individual WAV files for each slide. The tool supports both key-based and Microsoft Entra ID authentication and includes a dry-run mode for verifying SSML templates without incurring synthesis costs.

What it does

  • Extracts speaker notes from YAML files within a structured directory.
  • Generates WAV audio files using Azure AI Speech neural voices.
  • Applies custom pronunciation aliases for acronyms and technical terms.
  • Supports SSML template verification through a dry-run mode.
  • Allows configuration of speech rate and specific voice selection.
  • Enables targeted synthesis for individual slides or entire decks.

When to use it

  • When you need to generate consistent synthetic narration for slide-based presentations.
  • When you require precise control over the pronunciation of technical acronyms in audio.
  • When integrating automated voice-over generation into a slide content pipeline.
  • When you need to verify SSML output before committing to full audio synthesis.

Advantages

  • Automates the conversion of text-based speaker notes into audio assets.
  • Provides granular control over pronunciation using custom lexicon files.
  • Supports secure authentication via Microsoft Entra ID.
  • Offers flexibility to re-synthesize specific slides without processing the entire deck.

Usage examples

“Generate voice-over files for all slides in the content directory.”
The script processes all slide-NNN folders in the content directory and outputs corresponding WAV files to the voice-over folder.
“Verify the SSML templates for my slides without generating audio.”
The tool runs in dry-run mode, printing the generated SSML to the console for review without calling the Azure Speech service.
“Synthesize audio for slide 005 using a specific voice and rate.”
The tool generates a WAV file for slide-005 using the specified Azure neural voice and prosody rate settings.

Install

  1. Download the skill as a ZIP:

    https://mp-staging.autogineer.com/agent-skills/microsoft/tts-voiceover/download
  2. Unzip it into the cross-client skills folder of your project:

    mkdir -p .agents/skills && unzip tts-voiceover.zip -d .agents/skills

Documentation by Microsoft

TTS Voice Over Skill

Generates per-slide WAV voice-over files from YAML speaker_notes using Azure AI Speech neural voices with SSML pronunciation control.

Overview

This skill reads content.yaml files from a slide content directory (slide-NNN/content.yaml), extracts speaker_notes fields, applies acronym aliases for correct pronunciation of technical terms, and produces one WAV file per slide. Supports dry-run mode for SSML template verification without Azure credentials.

Synthesis uses Azure AI Speech neural voices, including the HD voices such as en-US-Andrew:DragonHDLatestNeural. HD voices are offered in a subset of Azure regions; check the Speech service regions before choosing SPEECH_REGION.

Narration produced by this skill is synthetic. Tell listeners that the voice is AI-generated, following Microsoft's disclosure design guidelines for synthetic voices. HVE demo-material CI does not synthesize speech; it generates silent videos.

Prerequisites

  • Azure Speech resource: see Azure AI Speech pricing for the free-tier allowance and HD voice rates.

  • Authentication: Microsoft Entra ID (SPEECH_RESOURCE_ID, recommended) or a resource key (SPEECH_KEY).

  • Region: SPEECH_REGION is required for synthesis and has no default. Dry-run mode does not need it.

  • Python 3.11+ with uv for virtual environment management.

  • Data handling note: Speaker-notes content is transmitted to the configured SPEECH_REGION for synthesis. For prebuilt neural voices, Microsoft states that neither the input text nor the output audio is stored in Microsoft logs; see Data, privacy, and security for text to speech. Operators must still set an approved region and avoid sending regulated or confidential narration.

Entra ID avoids storing a long-lived key. It requires a custom domain on the Speech resource and the Cognitive Services Speech User role for the signed-in identity. The script resolves the identity through DefaultAzureCredential, so an az login session, a managed identity, or a GitHub Actions OIDC login through azure/login all work.

export SPEECH_RESOURCE_ID="/subscriptions/.../Microsoft.CognitiveServices/accounts/your-resource"
export SPEECH_REGION="eastus"
Key-Based Auth
export SPEECH_KEY="your-speech-key"
export SPEECH_REGION="eastus"

When both SPEECH_KEY and SPEECH_RESOURCE_ID are set, the script warns and uses the key. Unset SPEECH_KEY to use Entra ID.

Install dependencies:

# run from this skill folder
uv sync

Quick Start

Verify SSML templates without generating audio:

uv run scripts/generate_voiceover.py --dry-run --content-dir path/to/content

Generate voice-over WAV files:

uv run scripts/generate_voiceover.py --content-dir path/to/content --output-dir voice-over

Use demo-video to combine the WAV files with approved frames or clips. Use hve-slides for HTML presentations and hve-demo-material for the HTML/frame/MP4/caption workflow. Audio embedding into presentation files is retired.

Parameters Reference

For different speakers, synthesize one voice's selected slides per invocation using repeated --slide <number> arguments. Only those canonical slide-NNN folders are read and their WAVs replaced; other speakers' outputs are untouched. Omit --slide to synthesize the whole deck.

generate_voiceover.py
ParameterTypeDefaultDescription
--dry-runflagfalsePrint SSML without generating audio
--voicestringen-US-Andrew:DragonHDLatestNeuralAzure AI Speech voice name
--ratestring+10%Speech prosody rate
--content-dirpathcontentPath to slide content directory
--output-dirpathvoice-overPath to WAV output directory
--slideintegerall slidesRepeat to synthesize only the selected canonical slide numbers
--lexiconpath(auto-detect)Custom acronyms.yaml path
--collapse-newlinesflagfalseCollapse newlines and whitespace runs in speaker notes into single spaces before synthesis
--verbose / -vflagfalseEnable verbose (DEBUG) logging output

Script Reference

Generate with custom voice and rate:

uv run scripts/generate_voiceover.py \
  --content-dir content \
  --output-dir voice-over \
  --voice "en-US-Jenny:DragonHDLatestNeural" \
  --rate "+5%"

Use a custom lexicon:

uv run scripts/generate_voiceover.py \
  --content-dir content \
  --lexicon custom-acronyms.yaml

Collapse newlines in speaker notes (recommended for block-scalar | notes, whose line breaks are otherwise spoken as pauses):

uv run scripts/generate_voiceover.py \
  --content-dir content \
  --collapse-newlines

Acronym Lexicon

The lexicon controls SSML <sub alias> replacements for acronyms and technical terms. Create an acronyms.yaml file:

acronyms:
  HVE-Core: "H V E Core"
  OWASP: "Oh wasp"
  SBOM: "S Bomb"
  SLSA: "Salsa"
  CI/CD: "C I C D"

Lexicon resolution order:

  1. Path specified via --lexicon argument.

  2. acronyms.yaml in the content directory.

  3. Built-in defaults covering common technical acronyms.

SSML Template

Each slide produces an SSML document:

<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis"
 xmlns:mstts="http://www.w3.org/2001/mstts" xml:lang="en-US">
  <voice name="en-US-Andrew:DragonHDLatestNeural">
    <prosody rate="+10%">
      Text with <sub alias="Oh wasp">OWASP</sub> aliases applied.
    </prosody>
  </voice>
</speak>

Content Directory Contract

This skill reads a slide content directory with this structure:

content/
├── slide-001/
│   └── content.yaml    # Must include speaker_notes: field
├── slide-002/
│   └── content.yaml
└── ...

Each content.yaml should contain a speaker_notes: field with the narration text. The generated WAV files are named slide-NNN.wav matching the directory names.

Troubleshooting

IssueSolution
Set SPEECH_KEY ... or SPEECH_RESOURCE_IDExport SPEECH_KEY (key auth) or SPEECH_RESOURCE_ID (Entra ID) with SPEECH_REGION.
SPEECH_REGION must be setExport SPEECH_REGION with an approved Azure region. Synthesis has no default region.
401 with Entra ID authVerify custom domain on the Speech resource and Cognitive Services Speech User role. RBAC propagation takes up to 5 minutes.
Empty WAV files or skipped slidesVerify speaker_notes: is present and non-empty in content.yaml.
Mispronounced acronymsAdd entries to acronyms.yaml with phonetic aliases.
azure-cognitiveservices-speech package is requiredRun uv sync in the skill directory.

Files

tests/fuzz_harness.py

# Copyright (c) 2026 Microsoft Corporation. All rights reserved.
# SPDX-License-Identifier: MIT
"""Polyglot fuzz harness for TTS voice-over skill modules.

Runs as a pytest test when Atheris is not installed (CI default).
Runs as an Atheris coverage-guided fuzz target when executed directly.
"""

from __future__ import annotations

import sys
import tempfile
import xml.sax.saxutils
from contextlib import suppress
from pathlib import Path

try:
    import atheris

    FUZZING = True
except ImportError:
    FUZZING = False

from generate_voiceover import (
    _DEFAULT_ACRONYMS,
    apply_acronym_aliases,
    load_acronyms,
    wrap_ssml,
)

# ---------------------------------------------------------------------------
# Fuzz targets — pure functions exercised by both modes
# ---------------------------------------------------------------------------


def fuzz_apply_acronym_aliases(data):
    """Fuzz apply_acronym_aliases with random text and the default acronym dict."""
    if not FUZZING:
        return
    fdp = atheris.FuzzedDataProvider(data)
    raw_text = fdp.ConsumeUnicodeNoSurrogates(500)
    text = xml.sax.saxutils.escape(raw_text)
    with suppress(ValueError, TypeError):
        apply_acronym_aliases(text, dict(_DEFAULT_ACRONYMS))


def fuzz_wrap_ssml(data):
    """Fuzz wrap_ssml with random text, voice, and rate strings."""
    if not FUZZING:
        return
    fdp = atheris.FuzzedDataProvider(data)
    raw_text = fdp.ConsumeUnicodeNoSurrogates(200)
    text = xml.sax.saxutils.escape(raw_text)
    voice = fdp.ConsumeUnicodeNoSurrogates(50)
    rate = fdp.ConsumeUnicodeNoSurrogates(10)
    with suppress(ValueError, TypeError):
        wrap_ssml(text, voice, rate)


def fuzz_load_acronyms(data):
    """Fuzz load_acronyms with random YAML content written to a temp file."""
    if not FUZZING:
        return
    fdp = atheris.FuzzedDataProvider(data)
    content = fdp.ConsumeUnicodeNoSurrogates(300)
    with tempfile.NamedTemporaryFile(
        mode="w", suffix=".yaml", delete=False, encoding="utf-8"
    ) as tmp:
        tmp.write(content)
        tmp_path = Path(tmp.name)
    try:
        with suppress(Exception):
            load_acronyms(tmp_path)
    finally:
        tmp_path.unlink(missing_ok=True)


FUZZ_TARGETS = [
    fuzz_apply_acronym_aliases,
    fuzz_wrap_ssml,
    fuzz_load_acronyms,
]


def fuzz_dispatch(data):
    """Route Atheris input to one of the registered fuzz targets."""
    if len(data) < 2:
        return
    idx = data[0] % len(FUZZ_TARGETS)
    FUZZ_TARGETS[idx](data[1:])


# ---------------------------------------------------------------------------
# pytest mode — property-based tests for the same targets
# ---------------------------------------------------------------------------


class TestFuzzApplyAcronymAliases:
    """Property tests for apply_acronym_aliases."""

    def test_given_known_acronym_when_applied_then_sub_element_inserted(self):
        result = apply_acronym_aliases("Check OWASP guidelines", _DEFAULT_ACRONYMS)
        assert "Oh wasp" in result
        assert "<sub" in result

    def test_given_text_without_acronyms_when_applied_then_text_unchanged(self):
        result = apply_acronym_aliases("no acronyms here", _DEFAULT_ACRONYMS)
        assert result == "no acronyms here"

    def test_given_empty_text_when_applied_then_returns_empty(self):
        result = apply_acronym_aliases("", _DEFAULT_ACRONYMS)
        assert result == ""

    def test_given_empty_acronyms_when_applied_then_text_unchanged(self):
        result = apply_acronym_aliases("OWASP test", {})
        assert result == "OWASP test"

    def test_given_multiple_acronyms_when_applied_then_all_replaced(self):
        result = apply_acronym_aliases("ISE uses MCP", _DEFAULT_ACRONYMS)
        assert "I S E" in result
        assert "M C P" in result


class TestFuzzWrapSsml:
    """Property tests for wrap_ssml."""

    def test_given_voice_name_when_wrapped_then_voice_tag_present(self):
        result = wrap_ssml("hello", "en-US-Jenny", "+5%")
        assert "en-US-Jenny" in result
        assert "<voice" in result

    def test_given_rate_when_wrapped_then_prosody_rate_present(self):
        result = wrap_ssml("hello", "en-US-Jenny", "+10%")
        assert "+10%" in result
        assert "<prosody" in result

    def test_given_text_when_wrapped_then_text_in_output(self):
        result = wrap_ssml("test content", "voice", "rate")
        assert "test content" in result

    def test_given_any_input_when_wrapped_then_speak_root_element(self):
        result = wrap_ssml("x", "v", "r")
        assert result.startswith("<speak")
        assert result.endswith("</speak>")


class TestFuzzLoadAcronyms:
    """Property tests for load_acronyms."""

    def test_given_nonexistent_file_when_loaded_then_returns_defaults(self):
        result = load_acronyms(Path("/nonexistent/acronyms.yaml"))
        assert result == _DEFAULT_ACRONYMS

    def test_given_valid_yaml_when_loaded_then_returns_custom_map(self, tmp_path):
        lexicon = tmp_path / "acronyms.yaml"
        lexicon.write_text('acronyms:\n  FOO: "bar"\n', encoding="utf-8")
        result = load_acronyms(lexicon)
        assert result == {"FOO": "bar"}

    def test_given_invalid_format_when_loaded_then_returns_defaults(self, tmp_path):
        lexicon = tmp_path / "acronyms.yaml"
        lexicon.write_text("acronyms: not-a-dict\n", encoding="utf-8")
        result = load_acronyms(lexicon)
        assert result == _DEFAULT_ACRONYMS

    def test_given_empty_file_when_loaded_then_returns_defaults(self, tmp_path):
        lexicon = tmp_path / "acronyms.yaml"
        lexicon.write_text("", encoding="utf-8")
        result = load_acronyms(lexicon)
        assert result == _DEFAULT_ACRONYMS


# ---------------------------------------------------------------------------
# Atheris entry point — only runs when executed directly with Atheris installed
# ---------------------------------------------------------------------------

if __name__ == "__main__" and FUZZING:
    atheris.instrument_all()
    atheris.Setup(sys.argv, fuzz_dispatch)
    atheris.Fuzz()

Requirements and compatibility

Written for: GitHub Copilot

Works with: Claude Code, Claude apps, OpenAI Codex, GitHub Copilot, Cursor, Gemini CLI, Kiro, OpenCode, Goose, Amp, Roo Code, JetBrains Junie, Factory Droid, Devin (ex-Windsurf), Mistral Vibe, OpenHands, Letta

“Works with” means the tool reads the open Agent Skills format; a skill written for one tool may use features others lack.

Needs: Needs Python

Security

What our scanner noticed· Low risk

  • Scripts that your agent may run

This skill includes scripts that your agent may run. Read them before you use it.

A static scan is not a guarantee. Skills are third-party content: review a skill before you let an agent use it.

Attribution and licence

Published by
Microsoft
Publisher
Microsoft (Official)
Source
microsoft/hve-core · .github/skills/experimental/tts-voiceover · commit f425e94
Licence
MIT License (redistribution allowed)
  • LICENSE (added by AutoGineer from LICENSE)
  • ATTRIBUTION.md (written by AutoGineer)
Synced

AutoGineer is not affiliated with or endorsed by Microsoft.

Download ZIP

The ZIP holds the skill’s files, its licence files and an ATTRIBUTION.md. Download tts-voiceover.zip