Skip to content

yotsuda/Speech

BranchesTags

Repository files navigation

Speech

PowerShell modules for text-to-speech (TTS) and speech-to-text (STT) across multiple providers.

.NET 8.0 PowerShell 7.4+ License: MIT

Providers

Module TTS STT Requires
Speech.Windows Offline SAPI Offline SAPI Windows 10/11
Speech.Azure 400+ neural voices Real-time streaming Azure Speech key
Speech.OpenAI 11 multilingual voices Whisper (batch) OpenAI API key
Speech.Google Standard/WaveNet/Neural2 Batch Google Cloud credential JSON
Speech.Amazon Neural/standard voices Real-time streaming AWS access key + secret key
Speech.Core — — (shared config, microphone, output device)

Platform Support

Cmdlet Windows Linux/macOS
Out-*Speech (all providers) Yes Yes
Read-AzureSpeech Yes Yes
Read-GoogleSpeech Yes Yes
Read-AmazonSpeech Yes Yes
Read-WindowsSpeech Yes No (SAPI)
Read-OpenAISpeech Yes No (NAudio WinMM)

Quick Start

# Windows — no setup needed
Out-WindowsSpeech "Hello, world!"

# Azure
Set-AzureSpeechConfig -Key "your-key" -Region "eastus"
Out-AzureSpeech "Hello" -Language en-US

# OpenAI
Set-OpenAISpeechConfig -Key "sk-..."
Out-OpenAISpeech "Hello" -Voice nova

# Google
Set-GoogleSpeechConfig -Credential "path/to/key.json"
Out-GoogleSpeech "Hello"

# Amazon
Set-AmazonSpeechConfig -AccessKey "AKIA..." -SecretKey "..." -Region "ap-northeast-1"
Out-AmazonSpeech "Hello" -Voice Joanna

# Speech recognition (all providers)
$text = Read-WindowsSpeech
$text = Read-AzureSpeech -Language ja-JP
$text = Read-OpenAISpeech -Language ja
$text = Read-GoogleSpeech -Language ja-JP
$text = Read-AmazonSpeech -Language ja-JP

Installation & Configuration

Install-PSResource Speech

With PowerShell.MCP, AI can configure everything for you:

Install-PSResource PowerShell.MCP
claude mcp add PowerShell -s user -- "$(Get-MCPProxyPath)"

Then just ask:

Install the Az module and help me create an Azure Speech resource.

Help me set up OpenAI Speech. I don't have an API key yet.

Guide me through setting up Google Cloud Speech.

Help me set up Amazon Polly with my AWS credentials.

Say 'Hello world' using Windows Speech.

Windows SAPI works offline with zero configuration — the quickest way to get started.

Settings are stored in ~/Documents/PowerShell/Modules/Speech/SpeechConfig.json. API keys are masked when displayed.

Get-SpeechConfig          # View all settings
Get-SpeechConfig -Path    # Get config file path
Provider setup

Azure Speech Services

# Get key: Azure Portal > Create "Speech" resource > Keys and Endpoint
# Free tier (F0): 0.5M chars TTS + 5h STT / month
Set-AzureSpeechConfig -Key "your-key" -Region "eastus"
Get-AzureSpeech -Locale ja
Set-AzureSpeechConfig -Voice "ja-JP-NanamiNeural"

OpenAI

# Get key: https://platform.openai.com/api-keys
Set-OpenAISpeechConfig -Key "sk-..."
Set-OpenAISpeechConfig -Voice nova -Model tts-1

Google Cloud

# Get credential: Google Cloud Console > IAM > Service Accounts > Create key (JSON)
Set-GoogleSpeechConfig -Credential "C:\path\to\service-account.json"
Get-GoogleSpeech -Language ja-JP
Set-GoogleSpeechConfig -Voice "ja-JP-Neural2-B"

Amazon Polly / Transcribe

# Get credentials: AWS Console > IAM > Users > Create access key
# Free tier: 5M chars TTS + 60 min STT / month (first 12 months)
Set-AmazonSpeechConfig -AccessKey "AKIA..." -SecretKey "..." -Region "ap-northeast-1"
Get-AmazonSpeech -Language ja-JP
Set-AmazonSpeechConfig -Voice "Mizuki"

Windows

# No API key needed. Add voices: Settings > Time & language > Speech
Get-WindowsSpeech
Set-WindowsSpeechConfig -Voice "Microsoft Haruka Desktop"
Common options

All Out-*Speech cmdlets accept pipeline input and share these patterns:

# Pipeline
"Line 1", "Line 2" | Out-AzureSpeech

# Output device selection (Tab completion available)
Out-AzureSpeech "Hello" -OutputDevice "Speakers (Realtek)"
Set-SpeechConfig -OutputDevice "Speakers (Realtek)"   # persist

# Microphone selection
Read-AzureSpeech -Microphone "Headset Microphone"
Set-SpeechConfig -Microphone "Headset Microphone"     # persist

# Parameter > config priority for all settings
Out-AzureSpeech "Hello" -Key "temp-key" -Region "westus"  # one-time override

AI Voice Conversation

With PowerShell.MCP configured, AI can speak and listen through your speakers and microphone:

Let's have a voice conversation in English.

When I type 't', start listening and respond by voice.

Find me a good English voice and play a sample.

Compatible MCP Clients

Any MCP-compatible client that supports PowerShell.MCP can use Speech modules:

Cmdlet Reference

Each provider has 4 cmdlets following a consistent pattern:

Verb Purpose Example
Out-*Speech Text-to-speech Out-AzureSpeech "Hello"
Read-*Speech Speech-to-text $text = Read-AzureSpeech
Get-*Speech List voices Get-AzureSpeech -Locale ja
Set-*SpeechConfig Configure provider Set-AzureSpeechConfig -Voice "..."

Plus shared cmdlets in Speech.Core: Get-SpeechConfig, Set-SpeechConfig, Get-Microphone, Test-Microphone.

Use Get-Help <cmdlet> -Full for detailed documentation.

All 24 cmdlets

Speech.Core — Shared configuration and audio devices

  • Get-SpeechConfig — Display current configuration (-Path for file location)
  • Set-SpeechConfig — Set common settings: -Rate, -Volume, -Language, -Microphone, -OutputDevice
  • Get-Microphone — List audio input devices
  • Test-Microphone — Test microphone input level

Speech.Azure — Azure Cognitive Services

  • Out-AzureSpeech — TTS with SSML prosody (-Rate, -Volume, -Pitch, -Language, -Voice)
  • Read-AzureSpeech — Real-time streaming STT (-Language, -Detailed)
  • Get-AzureSpeech — List 400+ neural voices (-Locale to filter)
  • Set-AzureSpeechConfig — Set -Key, -Region, -Voice, -Pitch

Speech.OpenAI — OpenAI Audio API

  • Out-OpenAISpeech — TTS with 11 voices (-Voice, -Model, -Speed)
  • Read-OpenAISpeech — Whisper batch STT (-Language, -Model)
  • Get-OpenAISpeech — List available voices
  • Set-OpenAISpeechConfig — Set -Key, -Voice, -Model, -STTModel

Speech.Google — Google Cloud Speech

  • Out-GoogleSpeech — TTS with Standard/WaveNet/Neural2 (-Voice, -Language, -Speed)
  • Read-GoogleSpeech — Batch STT (-Language)
  • Get-GoogleSpeech — List available voices (-Language to filter)
  • Set-GoogleSpeechConfig — Set -Voice, -Credential

Speech.Amazon — Amazon Polly / Transcribe

  • Out-AmazonSpeech — TTS with neural/standard voices (-Voice, -Language, -Rate)
  • Read-AmazonSpeech — Real-time streaming STT (-Language)
  • Get-AmazonSpeech — List available voices (-Language to filter)
  • Set-AmazonSpeechConfig — Set -AccessKey, -SecretKey, -Region, -Voice

Speech.Windows — Windows SAPI

  • Out-WindowsSpeech — Offline TTS (-Voice, -Rate, -Volume)
  • Read-WindowsSpeech — Offline STT (-Language, -Confidence, -Detailed)
  • Get-WindowsSpeech — List installed SAPI voices (-Culture to filter)
  • Set-WindowsSpeechConfig — Set -Voice

Tab Completion

Most parameters support Tab or Ctrl+Space completion. Voice and language lists are fetched from each provider's API and cached for the session.

Cmdlet Tab-completable Parameters
Out-WindowsSpeech -Voice, -OutputDevice
Out-AzureSpeech -Language, -Voice, -OutputDevice
Out-OpenAISpeech -Model, -Voice, -OutputDevice
Out-GoogleSpeech -Language, -Voice, -OutputDevice
Out-AmazonSpeech -Language, -Voice, -OutputDevice
Read-WindowsSpeech -Culture, -Microphone
Read-AzureSpeech -Language, -Microphone
Read-OpenAISpeech -Language, -Model, -Microphone
Read-GoogleSpeech -Language, -Microphone
Read-AmazonSpeech -Language, -Microphone
Get-WindowsSpeech -Culture
Get-AzureSpeech -Locale
Get-GoogleSpeech -Language
Get-AmazonSpeech -Language
Set-*SpeechConfig -Voice, -Microphone, -OutputDevice 
# Language narrows the voice list
Out-AzureSpeech "Hello" -Language <Tab> -Voice <Tab>
# → en-US-JennyNeural, en-US-GuyNeural, ...

Out-OpenAISpeech "Hello" -Voice <Tab>
# → alloy, ash, ballad, coral, echo, fable, nova, onyx, sage, shimmer, verse

Read-AzureSpeech -Language <Tab>
# → en-US, ja-JP, zh-CN, ...

Read-OpenAISpeech -Microphone <Tab>
# → Headset Microphone, Microphone Array, ...

Troubleshooting

Common issues

"key not configured" / "credential not configured" Run the provider's Set-*Config cmdlet. See Get-Help Set-AzureSpeechConfig -Full.

No microphone input

Get-Microphone       # List devices
Test-Microphone      # Check input level (> 30 = OK)

Windows STT not recognizing language Install language pack: Settings > Time & language > Language & region > Add language > "Speech" feature.

License

MIT

Third-party: NAudio (MIT), Azure Speech SDK (MIT), AWS SDK for .NET (Apache-2.0).

About

PowerShell modules for text-to-speech (TTS) and speech-to-text (STT) across multiple providers.

www.powershellgallery.com/packages/Speech

Topics

cli text-to-speech powershell azure tts openai speech-to-text stt voicevox

Resources

Readme

License

MIT license
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages