FIL_mic_analysis // REPORT // 2026-01-25 12:17

Multi-Microphone Conversation Analysis

Mic Scale: 8ch Analysis: NLP + Spatial Interactive: Enabled

Interactive Temporal Analysis (10min Interval)

Hover over the bars to see exact word counts per time slot.

Temporal Topic Summaries

2026-01-17 21:30 リズム, ジュース, マックス, ソフト, ポリ
2026-01-17 21:40 全部, アイデア, it, オーディオ, OF
2026-01-17 21:50 コード, 最初, ビルド, データ, フラック
2026-01-17 22:00 設定, フレーム, 自分, フレー, その後
2026-01-17 22:10 結局, オープン, アメージ, OSC, コーディング
2026-01-17 22:20 ファイル, 人間, スキル, CMEC, メイク
2026-01-17 22:30 Yeah, おなか, オープン, sorry, カメラ
2026-01-17 22:40 it, to, AI, about, the
2026-01-17 22:50 AI, クロード, 最近, コンテクスト, 記憶
2026-01-17 23:00 最初, AI, 使用, レンズ, 機能
2026-01-17 23:10 AI, 自分, ボケ, プログラム, お願い
2026-01-17 23:20 拍手, 音楽, 身体, デザイン, 大学
2026-01-17 23:30 ガウシャン, 大阪, 交通, 電車, BCC
2026-01-17 23:40 CG, 千万, 大阪, 交通, 電車
2026-01-17 23:50 小川, 電車, 大阪, 交通, パイ
2026-01-18 00:00 大丈夫, ノイズ, データ, CG, AI
2026-01-18 00:10 練習, 一緒, レンズ, 機械, 自体
2026-01-18 00:20 失敗, 仕事, マイアーツ, CG, アメリカ
2026-01-18 00:30 Okay, 勉強, 知恵, メンシーフェキューシリーズ, 自分
2026-01-18 00:40 大丈夫, ポケット, 終了, 来週, 大地

Spatial Activity Animation (Time Slider)

Visualizing microphone activity over time. Use the slider below to replay the conversation flow (5min intervals).

Spatial Heatmap Interactive

Gaussian-interpolated loudness map evolving over time. Shows the "center of gravity" of the conversation in the room.

Word Co-occurrence Network

Visualizing relationships between frequent words. Nodes are words; edges indicate they were spoken in the same connected context.

Semantic Space Evolution (Embeddings + UMAP)

2D Animation (Slider)

Mapping the conversation's "meaning" into 2D space. Dots represent 5-minute conversation blocks. Similar topics cluster together. Use the slider to see the trajectory of topics over time.

3D Space-Time Cube (Z = Time)

Visualizing the conversation trajectory in 3D. The vertical Z-axis represents Time (flowing upwards). You can rotate the 3D cube to see the path of the conversation through semantic clusters.

3D Spatial Audio Log (Conversation Worm)

3D visualization of the spatial audio environment. The Z-axis represents Time. The "Tube" shows the center of the conversation moving through the room over time, and scatter points represent individual loud events.

Microphone Characteristic Vocabulary (TF-IDF)

Unique words for each microphone channel. These words appear frequently in that specific mic but rarely in others, highlighting localized conversations.

1. Project Overview

This project analyzed the conversational atmosphere of an event using 8 microphones. The goal was to visualize "what words were spoken" and "how the conversation changed over time" based on quantitative data.

2. Methodology

  1. Preprocessing: Extracted Nouns, Verbs, and Adjectives from transcripts (SRT).
  2. Frequency Analysis: Calculated word occurrences across all microphones.
  3. Temporal Analysis: Aggregated word usage into 10-minute windows.
  4. Spatial Analysis: Calculated RMS Amplitude from WAV files.

3. Key Findings

3.1 単語の定量化

イベント全体で最も頻繁に出現した単語は以下の通りです。

Word Frequency

3.2 Visual Summary (Word Cloud)

An intuitive visualization of the conversation's vocabulary.

Word Cloud

3.3 Temporal Trends

The heatmap below shows how the usage of top words changed over time.

(See assets/images/temporal_heatmap.png for the detailed heatmap)

Temporal Heatmap

3.4 Spatial Conversation Distribution

Please refer to the "Spatial Activity Animation" section above for the interactive visualization of microphone activity based on audio volume.

4. Technical Details

Report generated by AI Assistant.