omniai-google 3.8.0 → 3.9.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f8c007176bac8a2c1f07b5fb231997b9fdef9480cc75242145bdc628cde9d954
-  data.tar.gz: 4fbb50ea8ee2c1fae8a091712a70a89f872320483d946d0351e70d383683c2e8
+  metadata.gz: 5683bc29149bb31082f782ed460483773e2a88537290909d04adc3fb5285aad4
+  data.tar.gz: dd2cdde8860b212f25a4c761e1e48eccb31aeb0cbaa58dc963d48386f2352c6c
 SHA512:
-  metadata.gz: 38c769edd5f51804ab9cd3c88505681d109d3c4e078ec00813cea2074fb53952c9caea2e55751d22ca35ad23d158e086ae87aab4790b090c85350dd75ad30211
-  data.tar.gz: 3c5a7eca99512ff6ed4e4d2f2c51422154302d0ea89d9c13238ee8680771f497ffc05998b6fcb849b1b7ea9f89c89523826d10df28fc64913a77ed909c2bad9b
+  metadata.gz: 1880ce5da42484fa6f28ba61d14f4ee1d11cb4f3bf2d2b7d772b663af9222743775618dfc1f02de7ac1f50ac952ca742374b039722930ae28c6b388d67c0eff8
+  data.tar.gz: 6e8d16e2b1686ae61810ece5e5fb6c07fa69a4b0dcd25063261edc54e8765e228e3e40bd9fdde8b56fffc7b5c467f1a4e9c83708bc48b2bcd47ecfb144b52f31

data/README.md

@@ -266,20 +266,32 @@ client.transcribe("phone_call.mp3", model: OmniAI::Google::Transcribe::Model::TE
 # For medical conversations
 client.transcribe("medical_interview.mp3", model: OmniAI::Google::Transcribe::Model::MEDICAL_CONVERSATION)
 
+# Latest generation multilingual model (recommended)
+client.transcribe("audio.mp3", model: OmniAI::Google::Transcribe::Model::CHIRP_3)
+
 # Other available models
 client.transcribe("audio.mp3", model: OmniAI::Google::Transcribe::Model::CHIRP_2) # Enhanced model
 client.transcribe("audio.mp3", model: OmniAI::Google::Transcribe::Model::CHIRP)   # Universal model
 ```
 
 **Available Model Constants:**
+- `OmniAI::Google::Transcribe::Model::CHIRP_3` - Latest-generation multilingual ASR model (recommended)
+- `OmniAI::Google::Transcribe::Model::CHIRP_2` - Enhanced universal model
+- `OmniAI::Google::Transcribe::Model::CHIRP` - Universal model
 - `OmniAI::Google::Transcribe::Model::LATEST_SHORT` - Optimized for audio < 60 seconds
 - `OmniAI::Google::Transcribe::Model::LATEST_LONG` - Optimized for long-form audio
+- `OmniAI::Google::Transcribe::Model::TELEPHONY` - For phone/telephony audio
 - `OmniAI::Google::Transcribe::Model::TELEPHONY_SHORT` - For short phone calls
-- `OmniAI::Google::Transcribe::Model::TELEPHONY_LONG` - For long phone calls  
+- `OmniAI::Google::Transcribe::Model::TELEPHONY_LONG` - For long phone calls
 - `OmniAI::Google::Transcribe::Model::MEDICAL_CONVERSATION` - For medical conversations
 - `OmniAI::Google::Transcribe::Model::MEDICAL_DICTATION` - For medical dictation
-- `OmniAI::Google::Transcribe::Model::CHIRP_2` - Enhanced universal model
-- `OmniAI::Google::Transcribe::Model::CHIRP` - Universal model
+
+> **Region note:** `CHIRP_3` is only served from the `us` and `eu` multi-region endpoints (not `global`,
+> and not zonal regions like `us-east4`). The provider maps the configured `location_id` to its
+> multi-region parent — any `us*` region resolves to `us`, any `eu`/`europe*` region resolves to `eu` —
+> and defaults to `us` when nothing is configured. This means a Vertex AI client configured with a zonal
+> `location_id` (e.g. `us-east4`) for Gemini will still route `CHIRP_3` correctly. `CHIRP_2` is always
+> routed to `us-central1`.
 
 #### Supported Formats
 

data/lib/omniai/google/transcribe.rb

@@ -12,10 +12,12 @@ module OmniAI
       include TranscribeHelpers
 
       module Model
+        CHIRP_3 = "chirp_3"
         CHIRP_2 = "chirp_2"
         CHIRP = "chirp"
         LATEST_LONG = "latest_long"
         LATEST_SHORT = "latest_short"
+        TELEPHONY = "telephony"
         TELEPHONY_LONG = "telephony_long"
         TELEPHONY_SHORT = "telephony_short"
         MEDICAL_CONVERSATION = "medical_conversation"
@@ -111,7 +113,7 @@ module OmniAI
         # Speech-to-Text API uses different endpoints for regional vs global
         endpoint = speech_endpoint
         speech_connection = HTTP.persistent(endpoint)
-          .timeout(connect: @client.timeout, write: @client.timeout, read: @client.timeout)
+          .timeout(**http_timeout_options)
           .accept(:json)
 
         # Add authentication if using credentials

data/lib/omniai/google/transcribe_helpers.rb

@@ -17,16 +17,44 @@ module OmniAI
         case @model
         when "chirp_2"
           "us-central1"
+        when "chirp_3"
+          chirp_3_location_id
         else
           @client.instance_variable_get(:@location_id) || "global"
         end
       end
 
+      # Chirp 3 is only served from the `us` and `eu` multi-region endpoints (not `global`, and
+      # not zonal regions like `us-east4`). A Vertex client typically configures a zonal
+      # `location_id` for Gemini, so map any configured region to its multi-region parent and
+      # default to `us`.
+      #
+      # @return [String] "us" or "eu"
+      def chirp_3_location_id
+        case @client.instance_variable_get(:@location_id)
+        when /\A(eu|europe)/i then "eu"
+        else "us"
+        end
+      end
+
       # @return [String]
       def speech_endpoint
         location_id == "global" ? "https://speech.googleapis.com" : "https://#{location_id}-speech.googleapis.com"
       end
 
+      # Normalizes the client timeout into keyword args for HTTP.rb's `.timeout`. The speech
+      # endpoints build their own connections, so (unlike the base client, which passes the
+      # value straight through) they must accept both a scalar and a per-operation Hash. A Hash
+      # is passed through untouched; a scalar (or nil) is wrapped per-operation as before.
+      #
+      # @return [Hash]
+      def http_timeout_options
+        timeout = @client.timeout
+        return timeout if timeout.is_a?(Hash)
+
+        { connect: timeout, write: timeout, read: timeout }
+      end
+
       # @return [Array<String>, nil]
       def language_codes
         case @language
@@ -184,7 +212,7 @@ module OmniAI
       def poll_operation!(operation_name)
         endpoint = speech_endpoint
         connection = HTTP.persistent(endpoint)
-          .timeout(connect: @client.timeout, write: @client.timeout, read: @client.timeout)
+          .timeout(**http_timeout_options)
           .accept(:json)
 
         # Add authentication if using credentials
@@ -222,7 +250,7 @@ module OmniAI
       def request_batch!
         endpoint = speech_endpoint
         connection = HTTP.persistent(endpoint)
-          .timeout(connect: @client.timeout, write: @client.timeout, read: @client.timeout)
+          .timeout(**http_timeout_options)
           .accept(:json)
 
         # Add authentication if using credentials

data/lib/omniai/google/version.rb

@@ -2,6 +2,6 @@
 
 module OmniAI
   module Google
-    VERSION = "3.8.0"
+    VERSION = "3.9.0"
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: omniai-google
 version: !ruby/object:Gem::Version
-  version: 3.8.0
+  version: 3.9.0
 platform: ruby
 authors:
 - Kevin Sylvestre