omniai 1.1.4 → 1.1.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 62558072fec9583fe2df59e34498e31b449fc551a7d262a21ad45c1aa8bef721
-  data.tar.gz: 2b410f4fdccea4fdae2d839c60f703cbf311c3ad9b43f4f87b862f719c5e2ca8
+  metadata.gz: a82c1b8f8724f5dd7f35239ca2454e4ec7631fa1b8b0544249a49f8b6c52962f
+  data.tar.gz: 85e9a9ee74f30ad4c4ba6ef8fbbaa88dc0d31fea6927774d5c7a421e2d38e341
 SHA512:
-  metadata.gz: 5d975d4d7767392a20cab800bbcc17c988ae18adf0d6c31c816e99ab6b33cbeb9fac319d6e15bdbd425da3e00d2843c28004223f24c3471c1e2c60c55428fba4
-  data.tar.gz: e9967f2302492db51c6af31c101a908bf19392d5eefe87829a2d146431ea6d17b6d3516493a92beeae780b75cfb0e61a7214ddfd52817df5199f6d1e9b36642e
+  metadata.gz: 42c6324488864562480cf898c8e087ac841d035468a0f4b4dbcd7c86f4239d61ca8361c80e8dcd06ddb9dc565c2e04faa047c2e1342bacf1ac8d2d3825d85b25
+  data.tar.gz: cbeef3a03122130ce8c2bf634700df52a2a44aa1c013e8d0f2862379940965326f693f4e87f2704d0dd0a197cddf136cc5cd440d4c36a5a573ba00910845d044

data/README.md

@@ -96,7 +96,7 @@ client.chat('Tell me a joke.', stream:)
 
 ### Transcribe
 
-Clients that support chat (e.g. OpenAI w/ "Whisper", etc) convert recordings to text via the following calls:
+Clients that support transcribe (e.g. OpenAI w/ "Whisper") convert recordings to text via the following calls:
 
 #### Transcriptions with Path
 
@@ -108,7 +108,30 @@ transcription.text # '...'
 #### Transcriptions with Files
 
 ```ruby
-file = File.open("example.ogg", "rb")
-transcription = client.transcribe(file)
-transcription.text # '...'
+File.open("example.ogg", "rb") do |file|
+  transcription = client.transcribe(file)
+  transcription.text # '...'
+end
+```
+
+### Speak
+
+Clients that support speak (e.g. OpenAI w/ "Whisper") convert text to recordings via the following calls:
+
+#### Speech with Stream
+
+```ruby
+File.open('example.ogg', 'wb') do |file|
+  client.speak('The quick brown fox jumps over a lazy dog.', voice: 'HAL') do |chunk|
+    file << chunk
+  end
+end
+```
+
+#### Speech with File
+
+```ruby
+tempfile = client.speak('The quick brown fox jumps over a lazy dog.', voice: 'HAL')
+tempfile.close
+tempfile.unlink
 ```

data/lib/omniai/chat.rb

@@ -56,7 +56,7 @@ module OmniAI
       @format = format
     end
 
-    # @raise [ExecutionError]
+    # @raise [HTTPError]
     def process!
       response = request!
       raise HTTPError, response.flush unless response.status.ok?

data/lib/omniai/client.rb

@@ -59,5 +59,26 @@ module OmniAI
     def transcribe(io, model:, language: nil, prompt: nil, temperature: nil, format: nil)
       raise NotImplementedError, "#{self.class.name}#speak undefined"
     end
+
+    # @raise [OmniAI::Error]
+    #
+    # @param input [String] required
+    # @param model [String] required
+    # @param voice [String] required
+    # @param speed [Float] optional
+    # @param format [String] optional (default "aac"):
+    #   - "aac"
+    #   - "mp3"
+    #   - "flac"
+    #   - "opus"
+    #   - "pcm"
+    #   - "wav"
+    #
+    # @yield [output] optional
+    #
+    # @return [Tempfile``]
+    def speak(input, model:, voice:, speed: nil, format: nil, &stream)
+      raise NotImplementedError, "#{self.class.name}#speak undefined"
+    end
   end
 end

data/lib/omniai/speak.rb

@@ -0,0 +1,138 @@
+# frozen_string_literal: true
+
+module OmniAI
+  # An abstract class that provides a consistent interface for processing speak requests.
+  #
+  # Usage:
+  #
+  #   class OmniAI::OpenAI::Speak < OmniAI::Speakw
+  #     module Model
+  #       WHISPER_1 = "whisper-1"
+  #     end
+  #
+  #     protected
+  #
+  #     # @return [Hash]
+  #     def payload
+  #       raise NotImplementedError, "#{self.class.name}#payload undefined"
+  #     end
+  #
+  #     # @return [String]
+  #     def path
+  #       raise NotImplementedError, "#{self.class.name}#path undefined"
+  #     end
+  #   end
+  #
+  #   client.transcribe(File.open("..."), model: "...", format: :json)
+  class Speak
+    module Format
+      AAC = 'aac'
+      FLAC = 'flac'
+      MP3 = 'mp3'
+      OPUS = 'opus'
+      PCM = 'pcm'
+      WAV = 'wav'
+    end
+
+    # @raise [HTTPError]
+    #
+    # @param client [OmniAI::Client] required
+    # @param input [String] required
+    # @param model [String] required
+    # @param voice [String] required
+    # @param speed [Float] optional
+    # @param format [String] optional (default "aac"):
+    #   - "aac"
+    #   - "mp3"
+    #   - "flac"
+    #   - "opus"
+    #   - "pcm"
+    #   - "wav"
+    #
+    # @yield [chunk]
+    #
+    # @return [Tempfile]
+    def self.process!(input, client:, model:, voice:, speed: nil, format: nil, &)
+      new(input, client:, model:, voice:, speed:, format:).process!(&)
+    end
+
+    # @param client [OmniAI::Client] required
+    # @param input [String] required
+    # @param model [String] required
+    # @param voice [String] required
+    # @param speed [Float] optional
+    # @param format [String] optional (default "aac"):
+    #   - "aac"
+    #   - "mp3"
+    #   - "flac"
+    #   - "opus"
+    #   - "pcm"
+    #   - "wav"
+    def initialize(input, client:, model:, voice:, speed: nil, format: nil)
+      @input = input
+      @client = client
+      @model = model
+      @voice = voice
+      @speed = speed
+      @format = format
+    end
+
+    # @raise [HTTPError]
+    #
+    # @yield [chunk]
+    #
+    # @return [Tempfile]
+    def process!(&block)
+      response = request!
+      raise HTTPError, response.flush unless response.status.ok?
+
+      if block
+        stream!(response:, &block)
+      else
+        fetch!(response:)
+      end
+    end
+
+    protected
+
+    # @param response [HTTP::Response]
+    #
+    # @yield [chunk]
+    def stream!(response:, &block)
+      response.body.each { |chunk| block.call(chunk) }
+    end
+
+    # @param response [HTTP::Response]
+    #
+    # @return [Tempfile]
+    def fetch!(response:)
+      tempfile = Tempfile.new(['', ".#{@format}"])
+      tempfile.binmode
+      response.body.each { |chunk| tempfile << chunk }
+      tempfile.rewind
+      tempfile
+    end
+
+    # @return [Hash]
+    def payload
+      {
+        model: @model,
+        voice: @voice,
+        input: @input,
+        speed: @speed,
+      }.compact
+    end
+
+    # @return [String]
+    def path
+      raise NotImplementedError, "#{self.class.name}#path undefined"
+    end
+
+    # @return [HTTP::Response]
+    def request!
+      @client
+        .connection
+        .post(path, json: payload)
+    end
+  end
+end

data/lib/omniai/transcribe.rb

@@ -113,8 +113,8 @@ module OmniAI
       @client = client
     end
 
+    # @raise [HTTPError]
     # @return [OmniAI::Transcribe::Transcription]
-    # @raise [ExecutionError]
     def process!
       response = request!
       raise HTTPError, response.flush unless response.status.ok?

data/lib/omniai/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module OmniAI
-  VERSION = '1.1.4'
+  VERSION = '1.1.5'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: omniai
 version: !ruby/object:Gem::Version
-  version: 1.1.4
+  version: 1.1.5
 platform: ruby
 authors:
 - Kevin Sylvestre
@@ -76,6 +76,7 @@ files:
 - lib/omniai/chat/usage.rb
 - lib/omniai/client.rb
 - lib/omniai/config.rb
+- lib/omniai/speak.rb
 - lib/omniai/transcribe.rb
 - lib/omniai/transcribe/transcription.rb
 - lib/omniai/version.rb