Voice Diary

Transcribe voice messages from Telegram and save them as diary entries.

How It Works

1. User sends a voice message to Telegram

2. You download the voice file using Telegram Bot API

3. Transcribe using SkillBoss STT (Whisper)

4. Save to diary app

Configuration

Diary App URL: https://skillboss-worker-t4i97sfz.heyboss.live

Diary API Key: diary-voice-2026-xiaoyin

SkillBoss API: https://api.heybossai.com/v1

SkillBoss Key: See ~/.openclaw/.env (SKILLBOSS_API_KEY)

Step 1: Download Voice from Telegram

When you receive a voice message, extract the file_id and download it:

Get file path from Telegram
curl "https://api.telegram.org/bot${TELEGRAM_BOT_TOKEN}/getFile?file_id=FILE_ID"
Download the file (ogg format)
curl -o /tmp/voice.ogg "https://api.telegram.org/file/bot${TELEGRAM_BOT_TOKEN}/FILE_PATH"

Step 2: Transcribe with SkillBoss

Transcribe using SkillBoss (openai/whisper-1)
curl -X POST "https://api.heybossai.com/v1/audio/transcriptions" \
  -H "Authorization: Bearer $SKILLBOSS_API_KEY" \
  -F "file=@/tmp/voice.ogg" \
  -F "model=openai/whisper-1"

Response: {"text": "transcribed text here..."}

Step 3: Save to Diary

curl -X POST "https://skillboss-worker-t4i97sfz.heyboss.live/api/entries/voice" \
  -H "Authorization: Bearer diary-voice-2026-xiaoyin" \
  -H "Content-Type: application/json" \
  -d '{
    "content": "TRANSCRIBED_TEXT_HERE",
    "source": "telegram-voice",
    "mood": "neutral"
  }'

Quick Reference

For voice messages in Telegram:

1. Extract voice.file_id from the message

2. Download via Telegram Bot API

3. POST to SkillBoss /v1/audio/transcriptions

4. POST transcription to diary /api/entries/voice

5. Reply with confirmation: "Saved to diary: [first 50 chars]..."

Mood Detection (Optional)

Based on transcription, you can detect mood:

• Contains "happy", "great", "excited" → mood: "happy"
• Contains "sad", "tired", "frustrated" → mood: "sad"
• Contains "stressed", "anxious", "worried" → mood: "anxious"
• Default → mood: "neutral"

Example Flow

User sends voice: "Today was a really productive day. I finished the project and feel great about it."

You should:

1. Download the voice file

2. Transcribe to text

3. Detect mood: "happy" (contains "great", "productive")

4. Save to diary with title "Voice Note: Today was a really productive day..."

5. Reply: "Saved to diary: 'Today was a really productive day. I finished the project and feel great about it.'"

Telegram Bot Token

From OpenClaw config: 8422821644:AAEz1FumDcGaLexFD0TCgmAnguMSJFTNrj8

Use this to download voice files via Telegram Bot API.

SkillBoss API Key

From env: sk-gAAAAABpduGan3Tzc8XqnLdujVO3DfqA8aiLgOBbWvJ5l_WYZD7yUYNp8dTL7BhK1InesVlEgfihkU3yY-8TNeU-x5GiVkP1mhsMLS1IjQg9ZlM1Hkb6rcmKj4pOK11Bf3Q653KpnAufLFMCH5i2_2HnZ1GTxmM8nA==