Try Hablará, by Fidpa (Forschungsinstitut für digitale Barrierefreiheit gGmbH), to analyze and transcribe conversations locally for privacy-conscious reflection. It performs offline speech-to-text with whisper.cpp and applies psychological and linguistic models to detect emotions, tonality, cognitive biases, and argumentation fallacies on transcripts and raw audio. Includes Nonviolent Communication checks, Four-Sides Model analysis, 16 fallacy types, plus multi-language detection across 15 languages. Designed for coaches, therapists, researchers, and self-reflection users who need private, scientifically grounded conversation analysis.

Hablará performs on-device transcription and layered psychological analysis

The app converts audio to text using whisper.cpp and runs model-based analyses locally. It applies established frameworks such as the Four-Sides Model, Transactional Analysis, and Cognitive Bias detection, and it flags 16 argumentation fallacy types. Emotion tracking maps valence and arousal alongside formal versus informal tonality. The feature set targets analytical workflows rather than casual note-taking, making export options important for further research.

Local processing preserves data privacy but depends on local model capacity

Hablará is 100 percent offline-capable, so speech-to-text and analyses run on the device by default. The developer implements OS-native encryption for API keys and stores recordings locally to protect data. The app supports local model runtimes such as Ollama and Apple Intelligence for on-device inference, while offering optional integrations with cloud providers like OpenAI, Anthropic, and Mistral for users who choose to offload work.

Designed for professional workflows, with research-oriented exports and language support

The app offers export to Markdown, TXT, PDF, HTML, and DOCX to fit common research and coaching workflows. Automatic language detection covers fifteen languages, which helps multilingual recordings. The developer is a non-profit research institute focused on digital accessibility, a background that explains the inclusion of scientific frameworks and privacy controls. Casual users may find the analysis depth more useful when paired with domain knowledge in communication or therapy.

A practical, privacy-first option best for professionals and researchers

Hablará suits privacy-minded coaches, therapists, and researchers who need model-backed conversation analysis and exportable results for study. Its strengths lie in on-device transcription and theory-based evaluation designed for analytical workflows. The main caveat is dependency on local model capacity, so users without adequate hardware may offload large batches to optional cloud integrations or run heavy analyses on more powerful machines.