whisper.cpp 0.3.0 → 0.3.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 7d83008a506279672a2d585d35f232940c6d81ee6175ea866baa9965684fa4a8
-  data.tar.gz: 6e79798baa580cbcd6cdfb886f197bb6a7c1533a1b9ad9d6fa1d1b15cd708028
+  metadata.gz: 3d1b91cc9ac6e7682429f261d3e9b283c4c34ef246a2e88d07f1c294d145d1f4
+  data.tar.gz: be2a0ff1c4eda542685c209e6c64b7f3f3f902c4353534747922ea89c42ef24d
 SHA512:
-  metadata.gz: f41063a2b14007e68f7084c5c1e19e513d7ff89648cb6cb2fe3dfdd8982c868d20b45639799b3d57d210a7cd6aabaf7d5cc23d2a03b821f5dffa2a827dadf4a8
-  data.tar.gz: 5c69521555fde8f6080a437a927af31741978fc66efca0e825f4be058739ebd988e7df218b7cd485773ff99466c57361e256e684a912042e959191764845f342
+  metadata.gz: c992abd9e3e9b48d6b31cfa6489b5b09489781d251c2d3ca24d4932fca579b7d19b7483a1d72aefe77ac0655d5d5af19708bf6342cf6e5d4922d3663b1244acc
+  data.tar.gz: 8890fa1e7653f8690c3686a7751e597958c06adfaa3a0b3a3e74c947920f1da2014ada7eebe4fd78e85cc2a74fe235b244c43817937072ffe3b7c75792567ec1

data/lib/whisper/model.rb

@@ -8,6 +8,7 @@ module Whisper
     def initialize(model_path)
       @model_path = model_path
       @ctx = nil
+      init_whisper_context
     end
 
     def transcribe_from_file(audio_file_path, format: 'plaintext', **params)
@@ -17,8 +18,6 @@ module Whisper
     end
 
     def transcribe_from_audio_data(audio_data, format: 'plaintext', **params)
-      init_whisper_context(params)
-
       # Prepare full params
       full_params = default_full_params(params)
 
@@ -49,19 +48,13 @@ module Whisper
 
     private
 
-    def init_whisper_context(params)
+    def init_whisper_context params = {}
       return unless @ctx.nil?
 
       ctx_params = Whisper.whisper_context_default_params
 
-      # Set user-provided context params
-      user_ctx_params = params.fetch(:context_params, {})
-      user_ctx_params.each do |key, value|
-        if ctx_params.members.include?(field)
-          ctx_params[key] = value
-        else
-          warn "Unknown context_param field: #{field}"
-        end
+      params.select{ |k, _| ctx_params.members.include? k }.each do |key, value|
+        ctx_params[key] = value
       end
       ctx_params[:gpu_device] = ENV['WHISPER_GPU']&.to_i || 0
 
@@ -70,19 +63,18 @@ module Whisper
       raise 'Failed to initialize Whisper model' if @ctx.null?
     end
 
-    def default_full_params(params)
+    def default_full_params params = {}
       # Get default full params
       strategy = params.fetch(:sampling_strategy, Whisper::WHISPER_SAMPLING_GREEDY)
       full_params = Whisper.whisper_full_default_params(strategy)
 
+      # Set translate to false to prevent translation to English
+      full_params[:translate] = false
+      full_params[:language]  = FFI::MemoryPointer.from_string 'auto'
+
       # Set user-provided full params
-      user_full_params = params.fetch(:full_params, {})
-      user_full_params.each do |key, value|
-        if full_params.members.include?(field)
-          full_params[key] = value
-        else
-          warn "Unknown full_param field: #{field}"
-        end
+      params.select{ |k, _| full_params.members.include? k }.each do |key, value|
+        full_params[key] = value
       end
 
       full_params

data/lib/whisper/version.rb

@@ -1,5 +1,5 @@
 module Whisper
 
-  VERSION = '0.3.0'
+  VERSION = '0.3.2'
 
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: whisper.cpp
 version: !ruby/object:Gem::Version
-  version: 0.3.0
+  version: 0.3.2
 platform: ruby
 authors:
 - Braulio Oliveira