GitHub - schappim/gospeak: CLI for text-to-speech using OpenAI/Deepgram/Elevenlabs

gospeak

A self-contained command-line tool for text-to-speech using OpenAI, ElevenLabs, or Deepgram TTS APIs. Written in Go with no external dependencies like ffmpeg - just a single binary.

Features

Multiple TTS providers: OpenAI, ElevenLabs, and Deepgram
No ffmpeg required - uses native Go audio libraries
Multiple voice options for each provider
Standard and HD quality models
Adjustable speech speed
Read from arguments or stdin (perfect for piping)
Save to MP3 or play directly
Cross-platform audio playback

Requirements

macOS, Linux, or Windows
API key for your chosen provider (OpenAI, ElevenLabs, or Deepgram)

Installation

Homebrew (Recommended)

brew tap schappim/gospeak
brew install gospeak

See the homebrew-gospeak tap for more details.

Download Binary

Download the latest release from the releases page.

Build from Source

# Clone the repository
git clone https://github.com/schappim/gospeak.git
cd gospeak

# Build
go build -o gospeak .

# Optional: Install to PATH
sudo cp gospeak /usr/local/bin/

Configuration

Set your API key(s) as environment variables:

# For OpenAI (default provider)
export OPENAI_API_KEY="your-openai-api-key"

# For ElevenLabs
export ELEVENLABS_API_KEY="your-elevenlabs-api-key"

# For Deepgram
export DEEPGRAM_API_KEY="your-deepgram-api-key"

Or pass the key directly with the --token flag.

Usage

Basic Usage (OpenAI)

# Speak text directly (uses OpenAI by default)
gospeak "Hello, world!"

# Pipe text from another command
echo "Hello from the command line" | gospeak

# Use with other tools
cat article.txt | gospeak

Using ElevenLabs

# Switch to ElevenLabs provider
gospeak -p elevenlabs "Hello from ElevenLabs"

# Use a specific ElevenLabs voice
gospeak -p elevenlabs -v josh "Hello with Josh's voice"

# Use a custom voice_id directly
gospeak -p elevenlabs -v "your-custom-voice-id" "Hello"

Choose a Voice

OpenAI voices: alloy (default), echo, fable, onyx, nova, shimmer

gospeak -v nova "Hello with the nova voice"
gospeak -v echo "This is the echo voice"

ElevenLabs voices: rachel (default), domi, bella, antoni, elli, josh, arnold, adam, sam, george, charlie, emily, lily, michael

gospeak -p elevenlabs -v rachel "Hello with Rachel's voice"
gospeak -p elevenlabs -v josh "Hello with Josh's voice"

You can also use any ElevenLabs voice_id directly:

gospeak -p elevenlabs -v "21m00Tcm4TlvDq8ikWAM" "Using voice ID directly"

Using Deepgram

# Switch to Deepgram provider
gospeak -p deepgram "Hello from Deepgram"

# Use a specific Deepgram voice
gospeak -p deepgram -v asteria "Hello with Asteria"
gospeak -p deepgram -v luna "Hello with Luna"

# Use Aura 2 voices
gospeak -p deepgram -v thalia "Hello with Thalia (Aura 2)"
gospeak -p deepgram -v apollo "Hello with Apollo (Aura 2)"

# Use a full model name directly
gospeak -p deepgram -v "aura-asteria-en" "Using model name directly"

Deepgram voices: asteria (default), luna, stella, athena, hera, orion, arcas, perseus, angus, orpheus, helios, zeus

Deepgram Aura 2 voices: thalia, andromeda, helena, jason, apollo, ares

Hear All Voices (OpenAI)

Demo all OpenAI voices with the same text:

gospeak --all "The quick brown fox jumps over the lazy dog"

Save to File

# Save to MP3
gospeak -o output.mp3 "Save this to a file"

# Save and play
gospeak -o output.mp3 -s "Save and speak at the same time"

Adjust Speed

OpenAI: Speed ranges from 0.25 (slow) to 4.0 (fast)

gospeak -x 0.5 "Speaking slowly"
gospeak -x 2.0 "Speaking faster"

ElevenLabs: Speed ranges from 0.7 to 1.2

gospeak -p elevenlabs -x 0.8 "Speaking a bit slower"
gospeak -p elevenlabs -x 1.2 "Speaking faster"

ElevenLabs Voice Settings

Fine-tune ElevenLabs voice output:

# Adjust stability (0.0-1.0, default: 0.5)
gospeak -p elevenlabs --stability 0.8 "More stable voice"

# Adjust similarity boost (0.0-1.0, default: 0.75)
gospeak -p elevenlabs --similarity 0.9 "Higher similarity to original voice"

Use Different Models

OpenAI:

# HD model (default)
gospeak -m tts-1-hd "High definition audio"

# Standard model (faster, lower quality)
gospeak -m tts-1 "Standard quality"

ElevenLabs:

# Multilingual v2 (default)
gospeak -p elevenlabs -m eleven_multilingual_v2 "Multilingual model"

# Turbo v2.5 (faster)
gospeak -p elevenlabs -m eleven_turbo_v2_5 "Turbo model"

Options

Option	Short	Description	Default
`--provider`	`-p`	TTS provider (`openai`, `elevenlabs`)	`openai`
`--voice`	`-v`	Voice to use	Provider-specific
`--model`	`-m`	Model to use	Provider-specific
`--output`	`-o`	Save audio to file	-
`--speed`	`-x`	Speech speed	`1.0`
`--speak`	`-s`	Play audio even when saving to file	`false`
`--token`	-	API key	From env var
`--all`	-	Speak with all voices (OpenAI only)	`false`
`--stability`	-	Voice stability (ElevenLabs only)	`0.5`
`--similarity`	-	Similarity boost (ElevenLabs only)	`0.75`
`--help`	`-h`	Show help message	-

Provider Comparison

Feature	OpenAI	ElevenLabs	Deepgram
Env var	`OPENAI_API_KEY`	`ELEVENLABS_API_KEY`	`DEEPGRAM_API_KEY`
Default voice	`alloy`	`rachel`	`asteria`
Default model	`tts-1-hd`	`eleven_multilingual_v2`	`aura-asteria-en`
Speed range	0.25 - 4.0	0.7 - 1.2	Not supported
Voice count	6 built-in	14 presets + custom	18 presets + custom
Custom voices	No	Yes (via voice_id)	Yes (via model name)

Scripting Examples

Read a file aloud

Speak command output

Generate audio files for multiple texts

for voice in alloy echo fable onyx nova shimmer; do
  gospeak -v $voice -o "${voice}_sample.mp3" "This is the ${voice} voice"
done

Use with LLM output

# Pipe output from an LLM CLI tool
llm "Tell me a joke" | gospeak -v nova

# Use ElevenLabs for more natural speech
llm "Tell me a story" | gospeak -p elevenlabs -v josh

# Use Deepgram
llm "Tell me a fact" | gospeak -p deepgram -v asteria

Compare providers

# Same text with all providers
gospeak "Hello world"
gospeak -p elevenlabs "Hello world"
gospeak -p deepgram "Hello world"

Error Handling

When an error occurs, the tool outputs a message to stderr:

Error: OPENAI_API_KEY environment variable not set and --token not provided
Error: ELEVENLABS_API_KEY environment variable not set and --token not provided
Error: DEEPGRAM_API_KEY environment variable not set and --token not provided
Error: Invalid provider 'invalid'. Use 'openai', 'elevenlabs', or 'deepgram'
Error: Speed must be between 0.25 and 4.0 for OpenAI
Error: Speed must be between 0.7 and 1.2 for ElevenLabs
Warning: Speed adjustment is not supported for Deepgram, ignoring