Transcribo

Siddesh Patil, Prasad Alai, Yuvraj Rathod

Department of Information Technology,
14
The Bombay Salesian Society’s
Don Bosco Institute of Technology, Mumbai- 400070
2024-2025

Abstract Method/ Architecture / Design Details Results

In modern organizations, manual note-taking
during board meetings often results in Enrollment Enrollment Extracting Speaker
inefficiencies and inaccuracies. Transcribo Audio Necessary Features Embedded Data
addresses this problem by providing a
comprehensive boardroom management system
with real-time transcription and speaker
diarization. Utilizing Whisper-Medium for
transcription and Picovoice's Falcon for speaker Pyannote + Falcon Speaker
segmentation, the system ensures accurate and Meeting Audio Feature
Identificati
efficient minutes-of-meeting generation. Extraction on
Additionally, speaker identification and labeling
using PyAnnote enhances clarity. The solution
Whisper
automates the transcription process, summarizes
discussions, and shares the minutes with
participants, streamlining decision-making and Transcript
boosting productivity.

Introduction Implementation Methods/Algorithm/Pseudocode References

In contemporary organizational environments, Whisper-Medium: [‌ 1] S. K. Gaikwad, B. W. Gawali, and P. Yannawar, ”A Review on Speech
clear and accurate meeting records are essential Performs Automatic Speech Recognition (ASR) to convert speech into text. Recognition Technique,” Dr. Babasaheb Ambedkar Marathwada
University, Aurangabad, 2023.‌‌
for effective decision-making. However, traditional Uses a transformer-based architecture to generate accurate transcriptions in real-time.
manual note-taking often leads to incomplete or [2] T. von Neumann, C. Boeddeker, T. Cord-Landwehr, M. Delcroix,
Picovoice's Falcon:
inaccurate minutes, causing misunderstandings Conducts Speaker Diarization by detecting speaker boundaries. and R. Haeb-Umbach, ”Meeting Recognition with Continuous Speech
and inefficiencies. Existing methods lack the Separation and Transcription-Supported Diarization,” 2024.‌
Analyzes voice characteristics to identify when the speaker changes.
advanced technology needed for efficient
[3] K. M. Lyu, R. Y. Lyu, and H. T. Chang, ”Real-time multilingual
documentation, particularly in speech-to-text PyAnnote: speech recognition and speaker diarization system based on Whisper
transcription and room booking management. Performs Speaker Identification and labels each speaker. segmentation,” PeerJ Computer Science, vol. 10, 2024.‌‌
Utilizes deep learning models trained on speaker embeddings to match and assign speaker identities.
Transcribo offers a solution by automating the
[4] L. E. Shafey, H. Soltau, and I. Shafran, ”Joint speech recognition
minutes of meetings, providing accurate speaker LLaMA 3.1B: and speaker diarization via sequence transduction,” arXiv preprint
recognition and automated transcription. Generates concise summaries of the transcribed meeting content. arXiv:1907.05337, 2019.‌‌
By simplifying the documentation process, Utilizes natural language understanding to extract key points and generate coherent summaries.
[5] D. Al-Fraihat, Y. Sharrab, F. Alzyoud, A. Qahmash, M. Tarawneh,
Transcribo minimizes administrative effort, and A. Maaita, ”Speech recognition utilizing deep learning: A
enhances collaboration, and ensures actionable SMTP:
systematic review of the latest developments,” Human-centric
Sends transcriptions and summaries to participants via email.
insights for more effective decision-making. Computing and Information Sciences, vol. 14, 2024.‌
WorkEstablishes a secure connection to the mail server and handles email transmission.