LOGO

Sonogram Visible Voice - Voice Spectrogram Software

August 26, 2010
Sonogram Visible Voice - Voice Spectrogram Software

Analyzing Voice Evidence: A Technological Approach

A recent narrative I was developing centered around an individual who was the subject of a recorded phone conversation. Subsequently, this person denied having ever participated in the call.

Possessing both the audio recording of the call and a segment of the individual’s denial, I began exploring methods to demonstrate the vocal consistency between the two samples.

A Personal Interest in Voice Technology

I have a strong interest in the field of voice technologies. This explains my anticipation for advancements in Google Voice’s voice recognition capabilities, and my enthusiasm for PC voice control applications such as Tazti.

However, when the task involved a direct digital comparison of voices, I initially found myself without a clear solution.

Popular culture often depicts sophisticated voice identification systems, as seen in spy films, where computers can instantly match voices to known individuals using voice prints.

The Power of Spectrograms

My research led me to Sonogram Visible Speech, and I discovered that spectrogram-based voice technology is, in fact, a currently applicable method for reliably identifying individuals based solely on their voice.

Spectrogram analysis provides a visual representation of voice characteristics, allowing for detailed comparison and verification.

This technology offers a robust means of establishing vocal identity, even in cases where individuals attempt to disavow their own speech.

Understanding Spectrograms

Similar to how chemists utilize chemical isotopes to determine the composition of substances by isolating elements, a spectrogram deconstructs audio into its fundamental frequencies. The human voice is complex; individuals don't speak at a single frequency. Rather, the unique characteristics of your mouth, nasal cavities, and vocal cords contribute to a distinctive blend of frequencies.

Sonogram Visible Speech is a freely available software application designed to analyze audio and video files. It dissects the audio track, revealing the complete spectrum of frequencies present throughout its duration. The resulting spectrogram is visually represented as shown below.

hand-voice-recognition-sonogram-visible-voice-tool-1.jpg

The lower portion of the display resembles a standard sound wave, similar to what you might observe in audio editing software like Audacity. However, the central pane illustrates the frequency distribution of each segment of the audio file. This software offers a variety of waveform options for in-depth sound file examination, catering to more experienced users.

hand-voice-recognition-sonogram-visible-voice-tool-2.jpg

The display settings for each waveform can be customized through the "Options" menu, specifically under "General Adjustment." Here, you can define the calculations used for logarithmic graphs and adjust the overall presentation of all available charts.

hand-voice-recognition-sonogram-visible-voice-tool-3.jpg

When analyzing quiet sounds, or whispers, consider utilizing the logarithmic frequency display. This feature is enabled via the "Options" menu by selecting "Logarithmic Frequency." It effectively amplifies the significant frequency areas within the spectrogram.

hand-voice-recognition-sonogram-visible-voice-tool-4.jpg

This enhancement can be particularly helpful in identifying distinct frequency patterns that characterize an individual's voice. If you are unfamiliar with spectrogram analysis, the comprehensive Sonogram Online Help manual is an excellent resource. Access it by clicking "Help" and then "Online Help."

hand-voice-recognition-sonogram-visible-voice-tool-5.jpg

This manual provides a thorough introduction to the software and the principles of audio analysis using spectrograms.

Spectrogram Analysis Applied to Ghost Hunting Investigations

This software’s versatility allows for application across numerous fields. A common phenomenon encountered during ghost hunting – a pursuit I personally find engaging – is known as "electronic voice phenomenon." This refers to the alleged appearance of voices, attributed to spirits or apparitions, within audio recordings.

Numerous such recordings are publicly available online. Consequently, I undertook to download several from websites dedicated to paranormal investigation and subject them to a detailed spectrogram analysis.

hand-voice-recognition-sonogram-visible-voice-tool-6.jpg

The spectrogram visualization demonstrates that the frequencies associated with the purported voice are generally situated within the lower spectrum. However, a more comprehensive understanding of the voices present in the recording necessitates examining the supplementary waveform displays.

The Autocorrelation View function calculates the "pitch" of the audio within the specific timeframe indicated by the mouse cursor’s position.

hand-voice-recognition-sonogram-visible-voice-tool-7.jpg

Analysis revealed that the "ghost's" voice exhibited an average pitch frequency of approximately 129.0 hz. When the recording was advanced to the segment featuring the investigator’s voice, the calculated pitch frequency rose to around 208.0 hz.

This difference is logical, given that the investigator is female, while the recorded "ghost" voice presents as male.

hand-voice-recognition-sonogram-visible-voice-tool-8.jpg

Further detail regarding the voices was uncovered by accessing the Fast Fourier display. This graphical representation rapidly decomposes the primary frequencies and presents them using a color-coded system.

hand-voice-recognition-sonogram-visible-voice-tool-9.jpg

In this instance, the frequency breakdown was relatively dispersed, encompassing both high and a significant number of low frequencies. Conversely, the investigator’s speech exhibited a more concentrated frequency pattern, leaning towards the higher end of the spectrum, as illustrated.

hand-voice-recognition-sonogram-visible-voice-tool-10.jpg

This preliminary analysis confirmed a distinct difference between the two voices. However, this represents merely a fundamental demonstration of the software’s extensive capabilities.

Essentially, any scenario where the decomposition of a sound wave’s frequencies proves beneficial can be effectively addressed with this tool. It is characterized by its ease of learning, rapid setup and configuration, and performance that equals or surpasses that of commercially available spectrogram software.

Are there projects you are working on that could benefit from spectrogram analysis? Have you previously experimented with Sonogram Visible Speech? Please share your experiences and insights in the comments section below.

#sonogram#voice spectrogram#spectrogram software#voice analysis#sound visualization#audio analysis