audio

OpenAI Audio Module

This Terraform module provides easy-to-use resources for OpenAI's audio capabilities, including:

• Speech-to-Text: Convert spoken language in audio files to written text
• Audio Transcription: Transcribe audio files with detailed information
• Audio Translation: Translate audio files from any language to English
• Text-to-Speech: Convert text to natural-sounding speech

Usage

module "audio" {
  source = "path/to/modules/audio"

  # Speech-to-Text Configuration
  enable_speech_to_text      = true
  speech_to_text_model       = "whisper-1"
  speech_to_text_file        = "path/to/audio/file.mp3"
  speech_to_text_language    = "en"  # Optional: ISO-639-1 format
  speech_to_text_prompt      = "Transcript of a meeting"  # Optional
  speech_to_text_temperature = 0     # Optional: 0-1 range

  # Audio Transcription Configuration 
  enable_audio_transcription      = true
  audio_transcription_model       = "whisper-1"
  audio_transcription_file        = "path/to/audio/file.mp3"
  audio_transcription_language    = "en"  # Optional: ISO-639-1 format
  audio_transcription_prompt      = "Transcript of a lecture"  # Optional
  audio_transcription_temperature = 0     # Optional: 0-1 range

  # Audio Translation Configuration
  enable_audio_translation      = true
  audio_translation_model       = "whisper-1"
  audio_translation_file        = "path/to/audio/file.mp3"
  audio_translation_prompt      = "Translation of a speech"  # Optional
  audio_translation_temperature = 0     # Optional: 0-1 range

  # Text-to-Speech Configuration
  enable_text_to_speech      = true
  text_to_speech_model       = "tts-1"  # Options: tts-1, tts-1-hd
  text_to_speech_input       = "Hello world! This is text being converted to speech."
  text_to_speech_voice       = "alloy"  # Options: alloy, echo, fable, onyx, nova, shimmer
  text_to_speech_speed       = 1.0      # Optional: 0.25-4.0 range
  text_to_speech_output_file = "path/to/output/speech.mp3"
}

Requirements

• Terraform >= 0.13.0
• OpenAI Provider

Input Variables

Speech-to-Text Variables

Name	Description	Type	Default	Required
enable_speech_to_text	Whether to enable speech-to-text conversion	bool	false	no
speech_to_text_model	Model to use for speech-to-text	string	"whisper-1"	no
speech_to_text_file	Path to the audio file to transcribe	string	""	yes (if enabled)
speech_to_text_language	Language of the input audio (ISO-639-1)	string	null	no
speech_to_text_prompt	Text to guide the model's style	string	null	no
speech_to_text_response_format	Format of transcript output	string	"json"	no
speech_to_text_temperature	Sampling temperature (0-1)	number	0	no

Audio Transcription Variables

Name	Description	Type	Default	Required
enable_audio_transcription	Whether to enable audio transcription	bool	false	no
audio_transcription_model	Model to use for transcription	string	"whisper-1"	no
audio_transcription_file	Path to the audio file to transcribe	string	""	yes (if enabled)
audio_transcription_language	Language of the input audio (ISO-639-1)	string	null	no
audio_transcription_prompt	Text to guide the model's style	string	null	no
audio_transcription_response_format	Format of transcript output	string	"json"	no
audio_transcription_temperature	Sampling temperature (0-1)	number	0	no

Audio Translation Variables

Name	Description	Type	Default	Required
enable_audio_translation	Whether to enable audio translation	bool	false	no
audio_translation_model	Model to use for translation	string	"whisper-1"	no
audio_translation_file	Path to the audio file to translate	string	""	yes (if enabled)
audio_translation_prompt	Text to guide the model's style	string	null	no
audio_translation_response_format	Format of translation output	string	"json"	no
audio_translation_temperature	Sampling temperature (0-1)	number	0	no

Text-to-Speech Variables

Name	Description	Type	Default	Required
enable_text_to_speech	Whether to enable text-to-speech	bool	false	no
text_to_speech_model	Model to use for text-to-speech	string	"tts-1"	no
text_to_speech_input	Text to convert to speech	string	""	yes (if enabled)
text_to_speech_voice	Voice to use for speech	string	"alloy"	no
text_to_speech_response_format	Format of audio output	string	"mp3"	no
text_to_speech_speed	Speed of audio output (0.25-4.0)	number	1.0	no
text_to_speech_output_file	Path to save the audio file	string	""	yes (if enabled)

Data Source Variables

Name	Description	Type	Default	Required
verify_tts_file	Whether to verify the TTS output file	bool	false	no
use_stt_data_source	Whether to use STT data source	bool	false	no
existing_stt_id	ID of existing STT resource to import	string	""	no
use_transcription_data_source	Whether to use transcription data source	bool	false	no
existing_transcription_id	ID of existing transcription to import	string	""	no
use_translation_data_source	Whether to use translation data source	bool	false	no
existing_translation_id	ID of existing translation to import	string	""	no

Outputs

Name	Description
speech_to_text_text	Transcribed text from speech-to-text
speech_to_text_created_at	Timestamp of speech-to-text generation
speech_to_text_id	ID of the speech-to-text resource
audio_transcription_text	Transcribed text from audio transcription
audio_transcription_duration	Duration of audio file in seconds
audio_transcription_id	ID of the audio transcription resource
audio_translation_text	Translated text from audio translation
audio_translation_duration	Duration of audio file in seconds
audio_translation_id	ID of the audio translation resource
text_to_speech_created_at	Timestamp of text-to-speech generation
text_to_speech_output_file	Path to the generated audio file
tts_file_exists	Whether the TTS file exists (from data source)
tts_file_size	Size of the TTS file in bytes (from data source)
tts_file_last_modified	Last modified timestamp of TTS file (from data source)
stt_data_source_id	ID from the STT data source
transcription_data_source_id	ID from the transcription data source
translation_data_source_id	ID from the translation data source

Using Data Sources

This module supports both creating new audio resources and importing existing ones:

module "audio" {
  source = "path/to/modules/audio"

  # Create and verify a text-to-speech file
  enable_text_to_speech    = true
  text_to_speech_input     = "Hello world!"
  text_to_speech_output_file = "./speech.mp3"
  verify_tts_file          = true  # Enable TTS data source
  
  # Import an existing speech-to-text resource
  use_stt_data_source      = true
  existing_stt_id          = "transcription-1234567890"
  
  # Create a new audio transcription and use data source
  enable_audio_transcription = true
  audio_transcription_file = "./speech.mp3"
  use_transcription_data_source = true
}

API Limitations

List Operations Not Supported

Important: The OpenAI API does not currently support listing operations for audio resources. The following variables and data sources are included in the module but are commented out in the implementation because the OpenAI API does not provide endpoints for these operations:

Name	Description	Status
retrieve_all_text_to_speech	Whether to retrieve all text-to-speech conversions	Not supported by API
filter_tts_by_model	Filter text-to-speech by model	Not supported by API
filter_tts_by_voice	Filter text-to-speech by voice	Not supported by API
retrieve_all_speech_to_text	Whether to retrieve all speech-to-text conversions	Not supported by API
filter_stt_by_model	Filter speech-to-text by model	Not supported by API
retrieve_all_transcriptions	Whether to retrieve all transcriptions	Not supported by API
filter_transcriptions_by_model	Filter transcriptions by model	Not supported by API
retrieve_all_translations	Whether to retrieve all translations	Not supported by API
filter_translations_by_model	Filter translations by model	Not supported by API

These limitations are imposed by the OpenAI API, not by the Terraform provider. If you need to manage multiple audio resources, consider implementing your own tracking system outside of the provider.

Supported Data Source Operations

The module fully supports retrieving individual audio resources using their specific IDs:

• openai_text_to_speech - Retrieve information about a specific text-to-speech file
• openai_speech_to_text - Retrieve a specific speech-to-text conversion by ID
• openai_audio_transcription - Retrieve a specific transcription by ID
• openai_audio_translation - Retrieve a specific translation by ID

For more examples, see the examples/audio directory in the provider repository.