feat(ee): Add transcription support for audio messages (#11670)

<img width="419" alt="Screenshot 2025-06-03 at 4 25 37 PM"
src="https://github.com/user-attachments/assets/4b6ddd11-9b91-4981-a571-83746cc4d40b"
/>


Fixes https://github.com/chatwoot/chatwoot/issues/10182

---------

Co-authored-by: Sojan Jose <sojan@pepalo.com>

enterprise/app/jobs/captain/conversation/response_builder_job.rb

@@ -49,10 +49,24 @@ class Captain::Conversation::ResponseBuilderJob < ApplicationJob
 
   def message_content(message)
     return message.content if message.content.present?
+    return 'User has shared a message without content' unless message.attachments.any?
 
-    return 'User has shared an attachment' if message.attachments.any?
+    audio_transcriptions = extract_audio_transcriptions(message.attachments)
+    return audio_transcriptions if audio_transcriptions.present?
 
-    'User has shared a message without content'
+    'User has shared an attachment'
+  end
+
+  def extract_audio_transcriptions(attachments)
+    audio_attachments = attachments.where(file_type: :audio)
+    return '' if audio_attachments.blank?
+
+    transcriptions = ''
+    audio_attachments.each do |attachment|
+      result = Messages::AudioTranscriptionService.new(attachment).perform
+      transcriptions += result[:transcriptions] if result[:success]
+    end
+    transcriptions
   end
 
   def determine_role(message)

enterprise/app/jobs/messages/audio_transcription_job.rb

@@ -0,0 +1,13 @@
+class Messages::AudioTranscriptionJob < ApplicationJob
+  queue_as :low
+
+  def perform(attachment_id)
+    attachment = Attachment.find_by(id: attachment_id)
+    return if attachment.blank?
+
+    Messages::AudioTranscriptionService.new(attachment).perform
+  rescue StandardError => e
+    Rails.logger.error "Error in AudioTranscriptionJob: #{e.message}"
+    ChatwootExceptionTracker.new(e).capture_exception
+  end
+end

enterprise/app/models/enterprise/concerns/attachment.rb

@@ -0,0 +1,15 @@
+module Enterprise::Concerns::Attachment
+  extend ActiveSupport::Concern
+
+  included do
+    after_create_commit :enqueue_audio_transcription
+  end
+
+  private
+
+  def enqueue_audio_transcription
+    return unless file_type.to_sym == :audio
+
+    Messages::AudioTranscriptionJob.perform_later(id)
+  end
+end

enterprise/app/services/messages/audio_transcription_service.rb

@@ -0,0 +1,67 @@
+class Messages::AudioTranscriptionService < Llm::BaseOpenAiService
+  attr_reader :attachment, :message, :account
+
+  def initialize(attachment)
+    super()
+    @attachment = attachment
+    @message = attachment.message
+    @account = message.account
+  end
+
+  def perform
+    return { error: 'Transcription limit exceeded' } unless can_transcribe?
+    return { error: 'Message not found' } if message.blank?
+
+    begin
+      transcriptions = transcribe_audio
+      Rails.logger.info "Audio transcription successful: #{transcriptions}"
+      { success: true, transcriptions: transcriptions }
+    rescue StandardError => e
+      ChatwootExceptionTracker.new(e).capture_exception
+      Rails.logger.error "Audio transcription failed: #{e.message}"
+      { error: "Transcription failed: #{e.message}" }
+    end
+  end
+
+  private
+
+  def can_transcribe?
+    account.audio_transcriptions.present? && account.usage_limits[:captain][:responses][:current_available].positive?
+  end
+
+  def fetch_audio_file
+    temp_dir = Rails.root.join('tmp/uploads')
+    FileUtils.mkdir_p(temp_dir)
+    temp_file_path = File.join(temp_dir, attachment.file.filename.to_s)
+    File.write(temp_file_path, attachment.file.download, mode: 'wb')
+    temp_file_path
+  end
+
+  def transcribe_audio
+    transcribed_text = attachment.meta&.[]('transcribed_text') || ''
+    return transcribed_text if transcribed_text.present?
+
+    temp_file_path = fetch_audio_file
+
+    response = @client.audio.transcribe(
+      parameters: {
+        model: 'whisper-1',
+        file: File.open(temp_file_path),
+        temperature: 0.4
+      }
+    )
+
+    FileUtils.rm_f(temp_file_path)
+
+    update_transcription(response['text'])
+    response['text']
+  end
+
+  def update_transcription(transcribed_text)
+    return if transcribed_text.blank?
+
+    attachment.update!(meta: { transcribed_text: transcribed_text })
+    message.reload.send_update_event
+    message.account.increment_response_usage
+  end
+end