Official Application – Tongyi Tingwu Agent Product Introduction
The Tongyi Tingwu Agent combines intelligent speech AI and the Qwen large language model (LLM) to deliver targeted audio and video analysis solutions for vertical industries. Use cases include enterprise office scenarios (such as meetings, interviews, and training), online education, sales and service insights, and automotive industry production–supply–sales workflows, including customer profiling for car buyers and industrial speech quality analysis. Powered by two engines—low-barrier invocation and high-accuracy processing—the solution helps enterprises quickly unlock business data value.
Application overview
- Automotive sales and service insights: Analyze end-to-end car sales processes—including phone outreach, in-store reception, test drives, and after-sales service—to assess service quality. The analysis covers dialogue topics such as service procedures, business requirements, competitive advantages, and competitor response strategies. The full workflow uses an automotive-domain-specific automatic speech recognition (ASR) model and an insight model fine-tuned with supervised fine-tuning (SFT). This ensures high transcription and insight accuracy. Supports multiple audio capture methods, including phone calls, mobile devices, employee ID badges, and in-vehicle microphones.
- Car buyer profiling: Build comprehensive user profiles based on customer questions during phone calls, in-store visits, and test drives. Profiles cover key concerns such as vehicle models and configurations, ride experience, ownership cost, competitor feedback, and order discounts. This supports sales operations and frontline staff in refining follow-up strategies and improving conversion rates. Uses the ccai-pro model—enhanced for semantic understanding and instruction following—to ensure high-quality profiling results.
- General service insights: Analyze service interactions across telesales, field marketing, in-store reception, after-sales support, customer service, and ticket handling. Covers dialogue topics such as service procedures, business requirements, competitive advantages, and competitor response strategies. The full workflow uses a general-purpose ASR model and an SFT-fine-tuned insight model, ensuring high transcription and insight accuracy. Supports phone calls, mobile devices, and employee ID badges.
- Industrial production instruction transcription: Transcribe voice commands used in industrial quality inspection or equipment operation. This leverages a full-chain LLM optimized for domain-specific terminology to improve transcription accuracy, freeing workers’ hands and boosting productivity.
- Smart meeting notes: Use speech recognition and large language models to analyze real-time or offline recordings from meetings, interviews, and classrooms. Features include chapter previews, LLM-powered summaries (full-text summary, speaker summary, key-point recap, mind map), key-point extraction (keywords, action items, scenario identification), PPT extraction and summarization, spoken-to-written conversion, and custom prompts.
Application features
Automotive sales and service insights
- End-to-end coverage of automotive sales: Assess service quality across all touchpoints—from phone outreach and in-store reception to test drives—ensuring consistent service standards.
- Identify standardized service procedures and greeting protocols: Help sales teams master communication techniques and continuously improve customer satisfaction and business outcomes.
- Verify accurate product explanations and deal-closing efforts: Help sales staff convert leads more efficiently, increase close rates, and grow revenue.
Car buyer profiling
- Pinpoint customer concerns accurately: Help automotive sales teams identify specific customer needs around vehicle features, configurations, and safety—enabling personalized sales strategies.
- Reveal price sensitivity and purchase motivation: Guide sales staff in designing more attractive pricing and discount plans.
- Understand customer expectations for after-sales service: Provide data-driven support to boost customer satisfaction and loyalty—and earn long-term trust through exceptional after-sales experiences.
General service insights
- Comprehensive service quality assurance: Ensure every customer interaction meets high standards—improving satisfaction and strengthening company reputation.
- Accurate baseline checks: Identify common issues and improvement opportunities, delivering actionable feedback to raise service quality.
- Support continuous training optimization: Use data to strengthen team capabilities, drive more efficient workflows, and boost sales performance.
Industrial production instruction transcription
- Designed for industrial environments: Deliver highly accurate transcription of voice commands—even in noisy settings—boosting productivity and operational precision.
- Support Chinese, English, and abbreviations: Handle up to 1,000 instruction statements, offering seamless support for multilingual and technical-term use cases.
- Deliver ultra-high-accuracy transcription: Reduce human error, optimize production workflows, and let your team focus on innovation and quality.
Smart meeting notes
- High-accuracy speech recognition: Transcribe real-time audio streams or audio/video files into text. Supports transcription and real-time bidirectional translation for Chinese, English, Cantonese, Japanese, Korean, German, French, and Russian. Outputs include paragraph, sentence, and word-level start/end timestamps—ideal for synchronized caption display.
- Full LLM support: Includes chapter previews, LLM-powered summaries (full-text summary, speaker summary, key-point recap, mind map), key-point extraction (keywords, action items, scenario identification), PPT extraction and summarization, spoken-to-written conversion, and custom prompts.
Benefits
Built-in business attributes
- Automotive sales and service insights: Includes over 70 sales and service checkpoints to comprehensively assess service quality.
- Car buyer profiling: Includes over 50 car buyer concerns and three summary-level assessments to fully evaluate purchase intent.
Complete Agent workflow
- No-code configuration: Configure business rules visually in the console.
- Instant effect validation: Preview results immediately. Adjust and compare configurations in the UI to ensure alignment with business goals.
- Auto-generated code: Generate full-stack code automatically from your business configuration—minimizing integration effort while maximizing effectiveness.
Default best practices and flexible customization
- Pre-optimized model chain: Combines an automotive-domain-optimized ASR model, qwen-plus, and the ccai-pro model—enhanced for semantic understanding and instruction following—in a cascaded architecture. Delivers optimal multi-model orchestration across the full pipeline.
- Open model selection: Choose from multiple ASR and analysis models to match your business preferences.
Billing information
Billing method
- Pay-as-you-go is currently supported. Keep your account balance sufficient to avoid overdue payments that may disrupt service.
Billing items and standard pricing
- Audio and video file input
Feature | Price | Notes |
|---|---|---|
Audio and video speech recognition | 0.6 CNY/hour | Charged based on original audio file duration. Smart meeting notes includes these complimentary features:
|
Service insights | 0.13 CNY/hour | Charged based on original audio file duration. Applies to the “service insights” feature in both automotive sales and service insights and general service insights applications—and to the “detailed customer concern analysis” feature in the car buyer profiling application. |
Deep profiling | ||
Basic profiling | 0.064 CNY/hour | Charged based on original audio file duration. Applies to the “test-drive interest score,” “sales lead score,” or “one-sentence customer profile” features in the car buyer profiling application. |
Smart meeting notes – LLM analysis | Select features individually. Each capability is billed separately based on processed audio/video duration. When using multiple capabilities or prompts, charges stack (e.g., keywords + chapter preview = 0.064 + 0.064 = 0.128 CNY/hour).
| |
Smart meeting notes – multimodal analysis | 0.64 CNY/hour | Optional feature: video PPT extraction + PPT summary. |
Smart meeting notes – translation | Real-time: 4 CNY/hour Offline: 0.5 CNY/hour | Optional features include the following:
|
Industrial production instruction transcription | 1.8 CNY/1,000 requests | Billed per actual usage count. |
- Plain text input
Feature | Price (CNY/1,000 requests) | Notes |
|---|---|---|
Service insights | 1.9 | Each run of “service insights” or “detailed customer concern analysis” counts input text tokens. Every 150 tokens equals one request. Partial batches count as one request (e.g., 400 tokens = 3 requests). |
Deep profiling | ||
Basic profiling | 0.95 | In the car buyer profiling application, each run of “test-drive interest score,” “sales lead score,” or “one-sentence customer profile” counts input text tokens. Every 150 tokens equals one request. Partial batches count as one request (e.g., 400 tokens = 3 requests). |
Smart meeting notes | In the smart meeting notes application, billing is based on input text volume. Every 150 tokens equals one request. Partial batches count as one request (e.g., 400 tokens = 3 requests). |