Helping therapists to be their best

Lyssn is a technology company focused on improving the quality of mental health and addiction therapy.

What we offer

Session Recording, Review, and Annotation

Lyssn makes it easy to record, store, and review therapy sessions. Sessions are securely recorded from any internet-connected device via web-browser or app. Therapists have a dashboard with their clients and sessions, and supervisors can easily review their supervisees’ sessions. In addition, the session review tools make it easy to create and share notes about the session, and time-linked comments about important moments.

Automatic Transcripts and Session Dynamics

Lyssn’s speech processing pipeline provides session transcripts via automatic speech recognition, designed and tuned for psychotherapy. The recording dashboard also features an interactive report that provides easy-to-understand summmaries of the interpersonal process, including how much a therapist talked compared to her client and how many questions the therapist asked. The report also highlights emotionally intense moments, which therapsists might want to review with their supervisors.

Evidence-Based Treatment Fidelty

Lyssn’s machine learning / AI pipeline takes the spoken language of therapy and evaluates it relative to specific fidelity benchmarks. Our prediction models automatically estimate the therapist’s empathy, collaboration, reflections, and questions, to provide detailed performance-based feedback on Motivational Interviewing (MI) and Cognitive Behavioral Therapy (CBT, in development).


What is said in therapy is sensitive and demands the highest security. Our platform enforces stringent security in multiple ways, from our HIPAA-compliant system, to encrypted data transfers and storage, and two-factor authentication.

How Lyssn Works

Speaker Identification

Speech Recognition

Lyssn uses speech recognition to recognize words and sentences. Our technology also detects emotional tone, offering richer cues to help evaluate sessions.


Speaker Identification

In psychotherapy, there is more than one speaker. Lyssn differentiates between therapists and clients (“who speaks when?”). We partition the transcript into speaker-labeled talk turns for easier analysis.



Lyssn uses state-of-the-art machine learning technology to identify key session content, including specific therapist interventions, discussion topics, and emotions.



Lyssn provides an interactive tool for reviewing and annotating sessions, including high-level session summaries and detailed information contained within session transcripts.

Why Lyssn?


Feedback is instant.

Until now, feedback on the quality of psychotherapy relied on burdening patients with questionnaires, or slow and labor-intensive human evaluations of session recordings. Lyssn utilizes state-of-the-art machine learning technology, trained on thousands of human-evaluated sessions, to process the content of your session in real time.


Training, supervision, quality assurance.

Lyssn performance-based feedback is designed for therapist's own reflection and growth, training in specific evidence-based practices and skills, and to deepen and facilitate the process of supervision. We are also developing technology to provide insight on the quality and content of psychotherapy directly to patients.


User-centered design

Learning from your work should be easy. Our designers work directly with therapists and trainers to develop an easy-to-use, elegant platform so that you receive feedback in a way you can understand.


Lyssn cares about your privacy.

What clients share with their therapists is private, and we take security seriously. Our infrastructure utilizes two-factor authentication and encrypted, HIPAA-compliant cloud servers.

Backed by Years of Scientific Research

Imel et al. (2015)

  • Imel, Zac
  • Steyvers, Mark
  • Atkins, David

Computational psychotherapy research: Scaling up the evaluation of patient provider interactions


Xiao et al. (2015)

  • Xiao, Bo
  • Imel, Zac
  • Georgiou, Panayiotis


"Rate my therapist": Automated detection of empathy in drug and alcohol counseling


Hirsch et al. (2017)

  • Hirsch, Tad
  • Merced, Kritzia
  • Imel, Zac


Designing Contestability: Interaction Design, Machine Learning, and Mental Health.

ACM Conference on Designing Interactive Systems

Narayanan et al. (2013)

  • Narayanan, Shrikanath
  • Georgiou, Panayiotis

Behavioral Signal Processing: Deriving Human Behavioral Informatics From Speech and Language

Proceedings of the IEEE

Tanana et al. (2016)

  • Tanana, Michael
  • Hallgren, Kevin
  • Imel, Zac


A comparison of natural language processing methods for automated coding of motivational interviewing

Journal of Substance Abuse Treatment

Contact Us

Dave Atkins

Chief Executive Officer

Tad Hirsch

Chief Design Officer

Shri Narayanan

Chief Science Officer

Zac Imel

Chief Innovation Officer

Michael Tanana

Chief Technology Officer

Grin Lord

V.P. of Clinical Innovation