home / skills / willsigmon / sigstack / assemblyai-expert

assemblyai-expert skill

safe

/plugins/voice-input/skills/assemblyai-expert

This skill helps you integrate AssemblyAI's affordable transcription with AI features like sentiment, PII redaction, and real-time streaming.

npx playbooks add skill willsigmon/sigstack --skill assemblyai-expert

Review the files below or copy the command above to add this skill to your agents.

Files (1)

SKILL.md

2.4 KB

---
name: AssemblyAI Expert
description: AssemblyAI - cheapest transcription, AI features, sentiment analysis, PII redaction
allowed-tools: Read, Edit, Bash, WebFetch
model: sonnet
---

# AssemblyAI Expert

The most affordable transcription API with AI superpowers.

## Pricing (2026)

- **$0.0025/minute** base rate (cheapest!)
- Speaker diarization: +$0.02/hr
- Sentiment analysis: +$0.02/hr
- PII redaction: +$0.02/hr
- Free credits for testing

## Key Advantages
- Cheapest per-minute rate
- AI features built-in (sentiment, topics, summaries)
- PII redaction for compliance
- Real-time streaming option
- HIPAA-compliant option available

## Quick Start

### Install
```bash
pip install assemblyai
```

### Basic Transcription
```python
import assemblyai as aai

aai.settings.api_key = "your-api-key"
transcriber = aai.Transcriber()

transcript = transcriber.transcribe("audio.mp3")
print(transcript.text)
```

### With Speaker Labels
```python
config = aai.TranscriptionConfig(speaker_labels=True)
transcript = transcriber.transcribe("audio.mp3", config=config)

for utterance in transcript.utterances:
    print(f"Speaker {utterance.speaker}: {utterance.text}")
```

### With AI Features
```python
config = aai.TranscriptionConfig(
    speaker_labels=True,
    sentiment_analysis=True,
    auto_chapters=True,
    entity_detection=True,
    summarization=True
)

transcript = transcriber.transcribe("audio.mp3", config=config)

# Summary
print(transcript.summary)

# Sentiment per segment
for sentiment in transcript.sentiment_analysis:
    print(f"{sentiment.sentiment}: {sentiment.text}")

# Auto chapters
for chapter in transcript.chapters:
    print(f"{chapter.headline}: {chapter.summary}")
```

### PII Redaction
```python
config = aai.TranscriptionConfig(
    redact_pii=True,
    redact_pii_policies=[
        aai.PIIRedactionPolicy.person_name,
        aai.PIIRedactionPolicy.credit_card_number,
        aai.PIIRedactionPolicy.ssn
    ]
)
```

## Real-Time Streaming
```python
def on_data(transcript):
    if transcript.text:
        print(transcript.text, end="", flush=True)

transcriber = aai.RealtimeTranscriber(
    on_data=on_data,
    sample_rate=16000
)

transcriber.connect()
# Stream audio...
```

## Best For
- Budget-conscious projects
- Content analysis (sentiment, topics)
- Compliance needs (PII redaction)
- Podcast summarization
- Meeting insights

Use when: Cheapest option needed, AI analysis of audio, PII compliance

Overview

This skill provides a compact, practical integration with AssemblyAI for ultra-low-cost transcription and built-in AI audio analysis. It focuses on reliable transcripts, speaker diarization, sentiment and entity detection, auto-chapters, and PII redaction for compliance. The skill highlights real-time streaming and HIPAA-capable workflows while keeping cost visibility up front.

How this skill works

The skill calls AssemblyAI’s transcription API to upload audio, request analysis features, and fetch structured results. You can enable speaker labels, sentiment analysis, entity/topic detection, summarization, and configurable PII redaction policies. For low-latency needs it supports a realtime streaming client that invokes a callback as partial transcripts arrive.

When to use it

When you need the lowest per-minute transcription cost
To extract sentiment, topics, summaries or chapters from audio
When speaker diarization (who spoke when) is required
For PII-sensitive audio that must be redacted before storage or display
To stream live audio for immediate transcription and captioning

Best practices

Batch uploads for large volumes to minimize latency and manage costs
Combine speaker_labels with timestamps for clear meeting or podcast transcripts
Enable only the AI features you need to reduce per-hour add-ons (sentiment, PII, diarization are per-hour extras)
Use PII redaction policies proactively for regulated environments and log redaction decisions
Verify sample_rate and audio quality for realtime streams to avoid transcription errors

Example use cases

Automated meeting notes with speaker-attributed summaries and sentiment overviews
Podcast production: cheap bulk transcriptions plus auto-chapters and episode summaries
Customer support QA: sentiment per segment and entity detection to flag risky interactions
Healthcare or legal intake where PII redaction and HIPAA options are required
Live captions for events using realtime streaming with incremental transcript callbacks

FAQ

How is pricing structured?

Base transcription is charged per minute at a very low rate, with additional per-hour add-ons for speaker diarization, sentiment analysis, and PII redaction. Test with free credits to estimate costs.

Can I redact names and sensitive numbers automatically?

Yes. The skill supports configurable PII redaction policies (names, SSNs, credit card numbers, etc.) so sensitive items are removed from transcripts.

Does it support real-time streaming?

Yes. A realtime transcriber streams audio and delivers incremental transcript data via a callback for low-latency applications.