議事録生成についていくつかの機能追加 (システム音声、言語の指定、直接入力、見栄えの変更) (#1176)

Co-authored-by: sugusugi <[email protected]>

Author: Sugi275

packages/cdk/lambda/startTranscription.ts

@@ -3,6 +3,7 @@ import { APIGatewayProxyEvent, APIGatewayProxyResult } from 'aws-lambda';
 import {
   TranscribeClient,
   StartTranscriptionJobCommand,
+  LanguageCode,
 } from '@aws-sdk/client-transcribe';
 import { StartTranscriptionRequest } from 'generative-ai-use-cases';
 
@@ -14,13 +15,14 @@ export const handler = async (
     const req: StartTranscriptionRequest = JSON.parse(event.body!);
     const userId = event.requestContext.authorizer!.claims.sub;
 
-    const { audioUrl, speakerLabel, maxSpeakers } = req;
+    const { audioUrl, speakerLabel, maxSpeakers, languageCode } = req;
 
     const uuid = uuidv4();
 
     const command = new StartTranscriptionJobCommand({
-      IdentifyLanguage: true,
-      LanguageOptions: ['ja-JP', 'en-US'],
+      IdentifyLanguage: !languageCode, // Enable auto-detection when no language specified
+      LanguageCode: languageCode ? (languageCode as LanguageCode) : undefined, // Use specified language when provided
+      LanguageOptions: !languageCode ? ['ja-JP', 'en-US'] : undefined, // Language candidates for auto-detection only
       Media: { MediaFileUri: audioUrl },
       TranscriptionJobName: uuid,
       Settings: {

packages/types/src/protocol.d.ts

@@ -169,6 +169,7 @@ export type StartTranscriptionRequest = {
   audioUrl: string;
   speakerLabel: boolean;
   maxSpeakers: number;
+  languageCode?: string;
 };
 
 export type StartTranscriptionResponse = {

packages/web/public/locales/translation/en.yaml

@@ -662,11 +662,19 @@ meetingMinutes:
   frequency_1min: Every 1 minute
   frequency_5min: Every 5 minutes
   generate: Generate
-  generate_minutes_header: Generate Minutes
+  generate_minutes_header: Generate Meeting Minutes
   generated_minutes: Generated Minutes
   generating: Generating minutes...
   generation_error: Failed to generate meeting minutes. Please try again.
   generation_success: Meeting minutes generated successfully
+  language: Language
+  language_auto: Auto Detect
+  language_chinese: Chinese
+  language_english: English
+  language_japanese: Japanese
+  language_korean: Korean
+  language_thai: Thai
+  language_vietnamese: Vietnamese
   last_generated: Last generated {{time}}
   minutes_placeholder: Minutes will appear here after generation
   model: Model
@@ -824,11 +832,17 @@ summarize:
   title: Summarize
 transcribe:
   detailed_parameters: Detailed parameters
+  direct_input: Direct Input
+  direct_input_instruction: Please enter the source text for meeting minutes directly in the "Transcript" area below.
+  direct_input_placeholder: Please enter the source text for meeting minutes here...
   file_upload: File Upload
   max_speakers: Max Speakers
   max_speakers_help: Maximum number of speakers to recognize
   mic_input: Microphone Input
   result_placeholder: Speech recognition results will be displayed here
+  screen_audio: Screen Audio
+  screen_audio_error: 'Screen Audio Error:'
+  screen_audio_notice: 'For Windows: Select "Entire Screen" tab and turn ON "Share system audio".<br/>For Mac: Select "Chrome Tab" and turn ON "Also share tab audio".<br/>Chrome and Edge work. Firefox does not work.'
   select_input_method: Please select from microphone input or file upload
   speaker_names: Speaker names (comma separated)
   speaker_recognition: Speaker recognition

packages/web/public/locales/translation/ja.yaml

@@ -523,11 +523,19 @@ meetingMinutes:
   frequency_1min: 1分ごと
   frequency_5min: 5分ごと
   generate: 生成
-  generate_minutes_header: 議事録生成
+  generate_minutes_header: 議事録を生成
   generated_minutes: 生成された議事録
   generating: 議事録生成中...
   generation_error: 議事録の生成に失敗しました。再試行してください。
   generation_success: 議事録が正常に生成されました
+  language: 言語
+  language_auto: 自動検出
+  language_chinese: 中国語
+  language_english: 英語
+  language_japanese: 日本語
+  language_korean: 韓国語
+  language_thai: タイ語
+  language_vietnamese: ベトナム語
   last_generated: 最後の生成時刻 {{time}}
   minutes_placeholder: 生成後にここに議事録が表示されます
   model: モデル
@@ -662,11 +670,17 @@ summarize:
   title: 要約
 transcribe:
   detailed_parameters: 詳細なパラメータ
+  direct_input: 直接入力
+  direct_input_instruction: 下の「文字起こし」エリアに議事録の元となるテキストを直接入力してください。
+  direct_input_placeholder: ここに議事録の元となるテキストを入力してください...
   file_upload: ファイルアップロード
   max_speakers: Max Speakers
   max_speakers_help: 認識する話者の最大数
   mic_input: マイク入力
   result_placeholder: 音声認識結果がここに表示されます
+  screen_audio: スクリーン音声
+  screen_audio_error: 'スクリーン音声エラー:'
+  screen_audio_notice: 'Windows の場合:「画面全体」タブより、「システム音声も共有する」を ON にしてください。<br/>Mac の場合:「Chrome Tab」から、「Also share tab audio」を ON にしてください。<br/>Chrome と Edge は動作します。Firefox は動作しません。'
   select_input_method: マイク入力 or ファイルアップロードから選択してください
   speaker_names: 話し手の名前（カンマ区切り）
   speaker_recognition: 話者認識

packages/web/public/locales/translation/th.yaml

@@ -567,6 +567,14 @@ meetingMinutes:
   generating: กำลังสร้างรายงานการประชุม...
   generation_error: ไม่สามารถสร้างรายงานการประชุมได้ กรุณาลองอีกครั้ง
   generation_success: สร้างรายงานการประชุมสำเร็จ
+  language: ภาษา
+  language_auto: ตรวจจับอัตโนมัติ
+  language_chinese: จีน
+  language_english: อังกฤษ
+  language_japanese: ญี่ปุ่น
+  language_korean: เกาหลี
+  language_thai: ไทย
+  language_vietnamese: เวียดนาม
   last_generated: สร้างล่าสุดเมื่อ {{time}}
   minutes_placeholder: รายงานการประชุมจะปรากฏที่นี่หลังจากสร้าง
   model: โมเดล
@@ -692,11 +700,17 @@ summarize:
   title: สรุป
 transcribe:
   detailed_parameters: ตั้งค่าขั้นสูงแบบกำหนดเอง
+  direct_input: ป้อนข้อมูลโดยตรง
+  direct_input_instruction: โปรดป้อนข้อความต้นฉบับสำหรับรายงานการประชุมโดยตรงในพื้นที่ "บันทึกการถอดความ" ด้านล่าง
+  direct_input_placeholder: โปรดป้อนข้อความต้นฉบับสำหรับรายงานการประชุมที่นี่...
   file_upload: อัปโหลดไฟล์
   max_speakers: จำนวนผู้พูดสูงสุด
   max_speakers_help: จำนวนผู้พูดสูงสุดในบทสนทนาที่ต้องการแยกแยะ
   mic_input: อินพุตไมโครโฟน
   result_placeholder: ข้อความที่แปลงจากเสียงของคุณ
+  screen_audio: เสียงหน้าจอ
+  screen_audio_error: 'ข้อผิดพลาดเสียงหน้าจอ:'
+  screen_audio_notice: 'สำหรับ Windows: เลือกแท็บ "หน้าจอทั้งหมด" และเปิด "แชร์เสียงระบบด้วย"<br/>สำหรับ Mac: เลือก "Chrome Tab" และเปิด "Also share tab audio"<br/>Chrome และ Edge ใช้งานได้ Firefox ใช้งานไม่ได้'
   select_input_method: โปรดเลือกจากอินพุตไมโครโฟนหรืออัปโหลดไฟล์
   speaker_names: ชื่อผู้พูด (คั่นด้วยเครื่องหมายจุลภาค)
   speaker_recognition: การแยกแยะผู้พูด

packages/web/public/locales/translation/vi.yaml

@@ -537,6 +537,41 @@ language:
   ko: Tiếng Hàn
   th: tiếng Thái
   zh: Tiếng Trung
+meetingMinutes:
+  auto_generate: Tự động tạo
+  custom_prompt: Lời nhắc tùy chỉnh
+  custom_prompt_placeholder: Nhập lời nhắc tùy chỉnh để tạo biên bản cuộc họp...
+  frequency: Tần suất tạo
+  frequency_10min: Mỗi 10 phút
+  frequency_1min: Mỗi phút
+  frequency_5min: Mỗi 5 phút
+  generate: Tạo
+  generate_minutes_header: Tạo biên bản
+  generated_minutes: Biên bản đã được tạo
+  generating: Đang tạo biên bản...
+  generation_error: Không thể tạo biên bản cuộc họp. Vui lòng thử lại.
+  generation_success: Tạo biên bản cuộc họp thành công
+  language: Ngôn ngữ
+  language_auto: Tự động phát hiện
+  language_chinese: Tiếng Trung
+  language_english: Tiếng Anh
+  language_japanese: Tiếng Nhật
+  language_korean: Tiếng Hàn
+  language_thai: Tiếng Thái
+  language_vietnamese: Tiếng Việt
+  last_generated: Tạo lần cuối {{time}}
+  minutes_placeholder: Biên bản sẽ xuất hiện ở đây sau khi tạo
+  model: Mô hình
+  next_generation_in: 'Tạo tiếp theo trong : '
+  record_transcribe: Ghi âm và chuyển đổi
+  speech_recognition: Nhận dạng giọng nói
+  style: Kiểu biên bản
+  style_custom: Tùy chỉnh
+  style_faq: FAQ
+  style_newspaper: Báo chí
+  style_transcription: Chuyển đổi
+  title: Trình tạo biên bản cuộc họp
+  transcript: Bản chuyển đổi
 model:
   parameters:
     reasoning: Reasoning
@@ -649,11 +684,17 @@ summarize:
   title: Tóm tắt
 transcribe:
   detailed_parameters: Tham số chi tiết
+  direct_input: Nhập trực tiếp
+  direct_input_instruction: Vui lòng nhập trực tiếp nội dung nguồn cho biên bản cuộc họp vào vùng "Bản chuyển đổi" bên dưới
+  direct_input_placeholder: Vui lòng nhập nội dung nguồn cho biên bản cuộc họp tại đây...
   file_upload: Tải lên tệp
   max_speakers: Max Speakers
   max_speakers_help: Số lượng người nói tối đa được nhận dạng
   mic_input: Đầu vào microphone
   result_placeholder: Kết quả nhận dạng giọng nói sẽ được hiển thị ở đây
+  screen_audio: Âm thanh màn hình
+  screen_audio_error: 'Lỗi âm thanh màn hình:'
+  screen_audio_notice: 'Đối với Windows: Chọn tab "Toàn bộ màn hình" và bật "Chia sẻ âm thanh hệ thống".<br/>Đối với Mac: Chọn "Chrome Tab" và bật "Also share tab audio".<br/>Chrome và Edge hoạt động. Firefox không hoạt động.'
   select_input_method: Vui lòng chọn từ đầu vào microphone hoặc tải lên tệp
   speaker_names: Tên người nói (phân cách bằng dấu phẩy)
   speaker_recognition: Nhận dạng người nói

packages/web/public/locales/translation/zh.yaml

@@ -457,6 +457,14 @@ meetingMinutes:
   generating: 正在生成会议纪要...
   generation_error: 会议纪要生成失败。请重试。
   generation_success: 会议纪要生成成功
+  language: 语言
+  language_auto: 自动检测
+  language_chinese: 中文
+  language_english: 英语
+  language_japanese: 日语
+  language_korean: 韩语
+  language_thai: 泰语
+  language_vietnamese: 越南语
   last_generated: 最后生成时间 {{time}}
   minutes_placeholder: 生成后会议纪要将显示在这里
   model: 模型
@@ -566,11 +574,17 @@ summarize:
   title: 摘要
 transcribe:
   detailed_parameters: 详细参数
+  direct_input: 直接输入
+  direct_input_instruction: 请在下方"转录文本"区域中直接输入会议记录的源文本
+  direct_input_placeholder: 请在此处输入会议记录的源文本...
   file_upload: 文件上传
   max_speakers: 最大发言人数
   max_speakers_help: 要识别的最大发言人数
   mic_input: 麦克风输入
   result_placeholder: 语音识别结果将显示在这里
+  screen_audio: 屏幕音频
+  screen_audio_error: 屏幕音频错误：
+  screen_audio_notice: 'Windows 的情况:选择「整个屏幕」选项卡，并启用「共享系统音频」。<br/>Mac 的情况:选择「Chrome Tab」，并启用「Also share tab audio」。<br/>Chrome 和 Edge 可以使用。Firefox 无法使用。'
   select_input_method: 请选择麦克风输入或文件上传
   speaker_names: 发言人姓名（逗号分隔）
   speaker_recognition: 说话人识别

packages/web/src/hooks/useMicrophone.ts

@@ -36,6 +36,9 @@ const useMicrophone = () => {
   const [recording, setRecording] = useState(false);
   const [rawTranscripts, setRawTranscripts] = useState<
     {
+      resultId: string;
+      startTime: number;
+      endTime: number;
       isPartial: boolean;
       transcripts: Transcript[];
     }[]
@@ -181,6 +184,10 @@ const useMicrophone = () => {
                 const tmp = update(prev, {
                   $push: [
                     {
+                      resultId:
+                        result.ResultId ?? `mic-${Date.now()}-${Math.random()}`,
+                      startTime: result.StartTime ?? 0,
+                      endTime: result.EndTime ?? 0,
                       isPartial: result.IsPartial ?? false,
                       transcripts,
                     },
@@ -195,6 +202,11 @@ const useMicrophone = () => {
                       prev.length - 1,
                       1,
                       {
+                        resultId:
+                          result.ResultId ??
+                          `mic-${Date.now()}-${Math.random()}`,
+                        startTime: result.StartTime ?? 0,
+                        endTime: result.EndTime ?? 0,
                         isPartial: result.IsPartial ?? false,
                         transcripts,
                       },
@@ -259,6 +271,7 @@ const useMicrophone = () => {
     recording,
     transcriptMic,
     clearTranscripts,
+    rawTranscripts,
   };
 };