AssemblyAI Integration for Speech Recognition AssemblyAI is a cloud platform with a rich set of post-processing tools on top of STT: auto-chapters, entity extraction, sentiment detection, PII editing. Convenient when you need not just a transcript, but also analytics on the recording. ### Unique features - Auto Chapters - automatic splitting of the recording into thematic chapters - Entity Detection - people, companies, addresses, dates without additional configuration - Sentiment Analysis - sentiment by sentences - IAB Categories - content classification according to the IAB taxonomy - LeMUR - built-in LLM for questions on the transcript ### Integration via SDK
import assemblyai as aai
aai.settings.api_key = ASSEMBLYAI_API_KEY
config = aai.TranscriptionConfig(
language_code="ru",
speaker_labels=True,
punctuate=True,
format_text=True,
sentiment_analysis=True,
auto_chapters=True,
entity_detection=True
)
transcriber = aai.Transcriber(config=config)
transcript = transcriber.transcribe("https://example.com/audio.mp3")
for chapter in transcript.chapters:
print(f"{chapter.start}ms - {chapter.end}ms: {chapter.headline}")
# Вопрос к записи через LeMUR
result = transcript.lemur.task(
"Выдели ключевые решения, принятые на встрече",
final_model=aai.LemurModel.claude3_haiku
)
```### Cost: Basic transcription: $0.0065/minute. Best Plan with Speaker Diarization: $0.012/minute. Nano (fast model): $0.002/minute. Russian language support: basic, WER 15–20%. For Russian-language tasks, the Whisper-based model within AssemblyAI is recommended. Integration: 1 day.