Transcribe Audio

Upload audio or video files and receive accurate transcriptions with speaker detection, word-level timestamps, and export in multiple formats.

What you are building

A transcription pipeline where you upload a file, the system processes it with speech recognition, and you get back timestamped text with speaker labels.

When to use this approach

• Meeting recordings, interviews, or podcasts
• Video subtitling (export as SRT or VTT)
• Audio content indexing and search
• Compliance recordings that need written records

What you need

• A WAYSCloud account
• An audio or video file (max 2 GB)

---

Step 1 — Create a transcription job

Dashboard

1. Open Services → AI & Machine Learning → Speech Intelligence in the dashboard
2. Click New Transcription
3. Upload your file
4. Select language (or leave on auto-detect)

API

Request a presigned upload URL:

curl -X POST https://api.wayscloud.services/v1/dashboard/transcript/jobs \
  -H "Authorization: Bearer $JWT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "filename": "meeting-2025-12-15.mp3",
    "content_type": "audio/mpeg",
    "file_size_bytes": 52428800,
    "language": "auto"
  }'

Response:

{
  "job_id": "job-abc123",
  "upload_url": "https://storage.wayscloud.services/uploads/job-abc123?X-Amz-Signature=...",
  "upload_expires_in": 3600
}

Upload the file to the presigned URL:

curl -X PUT "https://storage.wayscloud.services/uploads/job-abc123?X-Amz-Signature=..." \
  -H "Content-Type: audio/mpeg" \
  --data-binary @meeting-2025-12-15.mp3

Confirm the upload:

curl -X POST https://api.wayscloud.services/v1/dashboard/transcript/jobs/job-abc123/upload-complete \
  -H "Authorization: Bearer $JWT_TOKEN"

---

Step 2 — Wait for processing

The job moves through: queued → processing → completed.

Poll the job status:

curl https://api.wayscloud.services/v1/dashboard/transcript/jobs/job-abc123 \
  -H "Authorization: Bearer $JWT_TOKEN"

Response (completed):

{
  "job_id": "job-abc123",
  "status": "completed",
  "language_detected": "en",
  "duration_seconds": 3600,
  "word_count": 4523,
  "segments": [
    {
      "start_time": 0.0,
      "end_time": 5.5,
      "text": "Hello, welcome to the meeting.",
      "confidence": 0.98,
      "speaker_id": 1
    }
  ]
}

---

Step 3 — Export the result

Generate an export in your preferred format:

curl -X POST https://api.wayscloud.services/v1/dashboard/transcript/jobs/job-abc123/export \
  -H "Authorization: Bearer $JWT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"format": "srt"}'

Response:

{
  "artifact_id": "artifact-xyz",
  "format": "srt",
  "download_url": "https://storage.wayscloud.services/artifacts/artifact-xyz.srt"
}

Supported formats: txt (plain text), json (full data), srt (subtitles), vtt (web subtitles).

---

Step 4 — Next steps

• Analyze with AI — Feed transcriptions to the LLM API for summarization. See Run an LLM Request.
• Store results — Save transcription files in Object Storage. See Store Files.
• Automate — Use the API to build automated transcription pipelines

Related services

• Speech Intelligence — full service documentation
• LLM API — process transcriptions with language models
• Object Storage — store audio files and results

Related guides

• Authentication — API key setup

API reference

• Speech Intelligence API — all transcription endpoints