Q1. How do I commission MuScene with an audio forensics case?
For authentication of recorded speech data against signs of falsification or tampering, or speaker identification of unknown speaker voices against recorded known speaker voice, you may send us through e-mail a written statement of your case including the following information:
1. Client information: name, legal proxy, etc. 2. Contact information: phone, mailing address, e-mail, etc. 3. Type of service you are searching for: e.g., speaker identification, audio recording validation (evidence of data falsification, editing, or tampering), noise reduction, speech signal amplification, etc. 4. Transcript(s) of the audio file(s) to be examined. 5. Case highlight: e.g., introductory description of your case and digital data you submit, key feature(s) of your recorded audio data that you wish to focus on, and any special instruction, etc. 6. A description of digital data you will pass on to MuScene: e.g., number of files, name and hash value(s) of each individual file(s), individual file description if the file name is not content-suggestive, etc.
*Send your inquiry, power of attorney and digital data to techsupport@voice-forensics.com *A copy of your digital audio data to be analyzed must accompany for issuing a price quotation; MuScene will then send you the quotation with payment bank information. Depending on the nature of your case, e.g., if the purpose of audio data authentication is targeted for ensuing legal procedures, the OFFICIAL FORENSICS copy of digital audio data may have to be submitted to MuScene for certification upon confirmation of your commission of your audio forensics case with MuScene.
Q2: How does MuScene proceed audio recording authentication?
We handle authentication services in stages: Stage 1: Digital audio data scrutiny for signs of potential post-recording manipulation.
Upon completing audio data scrutiny, a brief, summarizing written statement of the overall outcome (findings) will be presented to the client. At this point the client can have a clear perspective for proceeding toward an Official Forensics Report with the examiner’s affidavit. This preliminary conclusion reached enables the examiner to evaluate the amount of work to put into the Official Forensics Report and corresponding charge, which is directly tied with the complexity of fabrication techniques involved in discovered potential post-recording manipulations.
The client and/or proxy is entitled to detailed oral explanation of the findings as stated in the written summarizing statement through web-meeting or other equivalent approaches at no extra charge.
Stage 2: If an Official Forensics Report of the findings is desired, a quotation for issuing such a document will be sent to the client. The Official Forensics Report with the examiner’s affidavit will produce three hard copies, two will be sent to the client and one archival copy kept by MuScene. Before printing, the client may request additional copies at an additional charge.
The pre-tax cost for the first stage processing is largely determined by the length and quality of the digital audio data per se. Most of our previous services fell in the range USD 1,599~2,999.
The pre-tax cost for issuing a formal forensics report ranges from USD 599 for an authentic audio file where no sign of tampering could be detected, upwards to USD 2,499 for moderately technically challenging tampered audio files. These cost estimations apply to reports written in American English. Alternative arrangement in the language used (traditional Chinese instead of English) for the official written report can be made through discussion with our administrative team.
For time-sensitive cases such as criminal investigations, internal corporate inquiries, or commercial espionage requiring immediate forensic examination, MuScene Voice Forensics Laboratory offers a 24/7 expedited service. However, the acceptance of such requests is subject to the complexity of the case and our laboratory's current capacity. [For expedited cases, we can potentially reduce the initial forensic examination (Stage 1) timeline to 3-7 days. Clients can expect to be informed of any significant results as soon as they are available to aid in their decision-making process.]
To maintain the objectivity and impartiality of our findings, expedited service is limited to the initial stages of the forensic examination. The preparation of the formal report will adhere to our established internal review procedures.
*The results and data from the initial forensic examination (Stage 1) will not be compromised by expedited service requests.
Q3: How does MuScene proceed noise reduction, record restoration, and speech signal enhancement?
We have many tools for various tasks of audio signal manipulation (improvement, but exclusive of fabrication). Each tool in our toolbox applies unique and proprietary AI algorithms, filter processing, or signal processing algorithms, each selected to best match the individual quality of our client's data and the desired output requirement. Whenever feasible, the treated output data will be fully synchronized with your input audio signal.
MuScene takes no position regarding the content of your audio information and we try our best not to manipulate specific AI parameters so that the globally adopted notion about signal vs. noise is preserved. To be more specific, these guiding principles set a strict demarcation between signal clean-up from data falsification. Several versions of your original signal after clean-up with different parameters are produced enabling our client to cross-compare each treated versions so that important audio information could be discerned and extracted by the client.
If the client determines to present the acoustically improved audio data in a litigation setting, MuScene will prepare related audio data with accompanying written documents in a professional manner and in compliance with the client’s local legal requirements.
Q4: How does MuScene proceed speaker identification?
In typical settings, the client need to provide two sets of audio/video recordings accompanied with corresponding transcripts (the transcripts may be postponed until the start of Stage 2), one of the voice in question and one of the known speaker to which the unknown voice is to be compared against. The acoustic quality of the two versions of recordings should be as high as possible. The best outcome (high confidence) can be achieved with the known speaker’s speech samples taken in our audio laboratory in a critically controlled acoustic environment with specially designed speech material so that the audio examiner can proceed ensuing identification by means of text-dependent comparison approaches that allows highest confidence in conclusion to be achieved. The cost for in-house voice sampling is USD 899, and the cost for ensuing voice property analysis and examination will decrease significantly/be deducted correspondingly due to derived streamlining in procedures.
We find the following 4-stage scheme for speaker ID examination optimizes the balance of expectancy and investment from the client’s point of view. This scheme breaks down an involved speaker ID analysis into 4 stages with all prices quoted pre-tax:
Stage 1: Preliminary evaluation of audio recording qualities, within 3 working days for USD 199. Tasks include, but not limited to, confirming that the recorded speeches had been uttered in a natural and voluntary manner, the quality of speech signal, sufficiency of spoken words for comparison, etc. If the quality of the speech data meets the requirements for further processing but the quantity is too scant for an effective and confident conclusion to be made, more speech data may be requested from the client.
The service cost for Stage 2 processing will be made at the end of this Stage.
Stage 2: When the speech data meet the minimum requirement for speaker ID (Stage 1), or an in-house known-speaker voice sampling is arranged, an official power of attorney is signed with payment for Stage 2 before in-depth speech data analyses kick off. Tasks include, but not limited to, confirming the accuracy of the transcripts against the speech signals, picking words and phrases appropriate for acoustic/spectrographic analyses, normalization of acoustic parameters in the recorded signals, statistical analyses of basic speaker voices, etc.
This stage may take as many as 21 working days with costs for most previous cases falling in the range USD 1,599~3,999. The main goal to be achieved at this stage is a thorough auditory and linguistic analyses of the two sets of speakers that allows the examiner make a preliminary expert evaluation of the amount of instrumental acoustic analyses required in Stage 3 to achieve >75% confidence in the final conclusion making step. Only with thorough subjective knowledge of the speech data in mind can the examiner make a candid cost estimation based on the amount of ensuing work needed in Stage 3 to collect sufficient amount of scientific data required on which the conclusion is based.
In case that speech data turns out to be insufficient for an uncompromised conclusion with high confidence to be made, the client need to either provide incremental quantities of speech recordings or accept a final conclusion at reduced confidence level, or terminate the case without proceeding into Stage 3. At the conclusion of Stage 2, the client will be given a quotation for proceeding into Stage 3 along with an preliminary, unofficial prospective opinion of the examiner regarding the likely final conclusion achievable.
Stage 3: At the start of this stage, the examiner has formed his/her subjective opinion based on auditory scrutiny of the two sets of recorded speech data. The tasks in this stage are collecting supportive scientific evidence to validate the examiner’s subjective opinion formed at the conclusion of Stage 2. This stage typically takes 15~30 working days at a typical pre-tax charge of USD 1,699~4,699.
At the conclusion of Stage 3, a live discussion session will be arranged with the client accompanied with his/her legal representative for an in-depth explanation of all findings collected. This explanatory arrangement is provided at no additional charge. This live discussion session provides the client an opportunity to set ensuing litigation strategy (if applicable) as well as a bird’s eye view of the contents to be included in the official forensics report to be presented in Stage 4.
Stage 4: This is when the examiner composes the Official Forensics Report, which typically takes 15 ~ 30 working days at a charge of USD 999~2,999. The official report will be produced in triplet of which one copy will be withheld by MuScene as archive. Additional copies may be requested before printing at a small handling charge.
For time-sensitive cases such as criminal investigations, internal corporate inquiries, or commercial espionage requiring immediate forensic examination, MuScene Voice Forensics Laboratory offers a 24/7 expedited service. However, the acceptance of such requests is subject to the complexity of the case and our laboratory's current capacity. [For expedited cases, we can potentially reduce the overall forensic examination timeline (Stage 1 to Stage 3) by 40% to 65%. Clients can expect to be informed of any significant results as soon as they are available to aid in their decision-making process.]
To maintain the objectivity and impartiality of our findings, expedited service is limited to the Stage 1~3 of the forensic examination. The preparation of the formal report will adhere to our established internal review procedures.
*The results and data from the initial forensic examination (Stage 1~3) will not be compromised by expedited service requests.
Q5. How do you secure client's confidential data?
The MuScene Voice Forensics Laboratory holds full responsibility for securing our clients’ confidential information.
As a general practice, the Laboratory will keep the original digital data with any supplementary information for three months after the close of a commission all digital We withhold only one hardcopy of formal Forensic Report as our permanent official track record. As a result, any detailed intermediate forensic data provided to our clients during the investigation may not be retrievable later on from the Laboratory; and the clients are expected to hold their share of responsibility in keeping detailed processed data generated during analyzing the commissioned digital data.
Once the power of attorney has been filed only person(s) with name(s) listed in the document will be given access to ensuing communication regarding the proceedings of the case. Any third party entrusted by the client must present due evidence before access to confidential information of the investigation will be allowed. We pledged to uphold our clients’ trust in keeping their data and interpretation confidential. However, we preserve the right for autonomous revealing related information against the client in case the client should have publicly misinterpreted our examination of data; such autonomous disclosure of partial information only happens when the client knowingly distorted the conclusions reached by our forensic examiners. Under no other circumstance will the MuScene Voice Forensics Laboratory disclose information without the consent of the client, nor will we autonomously respond to unauthorized disclosure or conjecture of information by any unauthorized third party.
In short, as a general principle: the MuScene Voice Forensics Laboratory will not respond to unauthorized private, presumptive discussions of forensic audio examination scenarios in any form of self-media or social network. Exceptions may occur ➊ per our client’s request to underpin the facts (s)he proclaims from the examiner’s professional point of view, ➋ misinformation of our forensic findings have been widely spread that the forensic fact is on the brink of being subverted.
Our clients are welcome to draft their versions of confidentiality agreement in compliance with their interest that further binds the Laboratory over the commission.
Q6. My files, including documents, videos, and data, are too large to be sent via email. I'm unfamiliar with cloud storage and sending them via disc or external hard drive is both time-consuming and costly. Are there any alternative solutions? To submit your files (documents, images, data, digital evidence, commission letters, etc.), please request a cloud storage upload link from our laboratory staff. No account or password is required. For efficient processing, please ensure your files are clearly named and categorized. A detailed file list specifying the purpose or type of each file would be helpful. Kindly inform us upon completion of the upload. Thank you.
Q7. MuScene Voice Forensics Laboratory has previously collaborated with courts, law enforcement agencies, and private companies. If a client whose case I am entrusting to you has a previous cooperative relationship with your laboratory, will this affect the impartiality and independence of the case? Can the laboratory guarantee that the results of the forensic examination will not be influenced by any external factors?
MuScene Voice Forensics Laboratory, a privately funded third-party forensic laboratory and research organization, has long been entrusted with forensic investigations, collaborative research, and outsourced research projects by domestic and international government agencies, private organizations, and individuals.
MuScene maintains a strict third-party, unbiased protocol in all forensic cases, adhering to a rigorous anti-fraud mechanism (verification process and peer review system). Case assignments are handled through a random draw system, team-based approach, and a three-tiered independent review process (involving at least four individuals).
The laboratory strictly refuses to advocate for or against any specific party or position. Regardless of whether the client is a law enforcement agency, corporation, law firm, non-profit organization, research partner, sponsor, or marginalized group, MuScene only accepts commissions based on a neutral third-party standpoint. The data, findings, and expert conclusions derived from our analysis and investigations are not subject to any form of modification, textual adjustments, or alterations in tone or wording.
Q8. Regarding the determination of whether a digital file has been edited or tampered with, what specific criteria does MuScene Voice Forensics Laboratory use to assess the continuity, authenticity and evidentiary value of audio and video files?
Our Forensic Standards MuScene Voice Forensics Laboratory adheres to rigorous standards to ensure the accuracy and reliability of our forensic analyses. Our primary areas of examination include:
A. Integrity and Traceability of Digital Evidence: We meticulously examine the digital evidence chain to ensure its completeness and traceability (chain of custody). Any breaks or manipulations in the evidence chain can significantly impact the admissibility and weight of the evidence. To safeguard the rights of all parties involved, we thoroughly investigate potential alterations or tampering, considering all possible techniques and tools that might have been used to exploit such vulnerabilities.
B. Analysis of Audio and Video File Specifications, Parameters, and Encoding: We conduct a comprehensive analysis of the specifications, parameters, and encoding used to create the audio and video files. By comparing these attributes to our extensive database of known samples, we can identify any anomalies or inconsistencies that may indicate tampering or manipulation. Normal recordings should exhibit consistent specifications and parameters. Any deviations from these norms warrant further investigation.
C. Continuity of Audio and Video Recordings: Building upon the findings from sections A and B, we conduct a detailed analysis of the continuity of the audio and video recordings to arrive at preliminary conclusions.
D. Proprietary Forensic Techniques: To prevent the unauthorized use and exploitation of our proprietary forensic techniques, certain examination methods and criteria are considered confidential. These methods are only disclosed in written forensic reports when necessary.
Limitations: Please note that our laboratory does not have investigative powers. Therefore, to fully address certain inquiries, we may require additional information or investigative findings provided by the client. This information will be used in conjunction with our forensic analysis to draw more definitive conclusions.
Q9. Does MuScene Voice Forensics Laboratory's report have evidentiary value? How should I read and use the report? MuScene Voice Forensics Laboratory, a subsidiary of the American company MuScene, is dedicated to audio forensics and software development. We are an independent entity, unaffiliated with any government, political party, or region. Our funding sources are entirely independent of China, Hong Kong, and Taiwan. To ensure the scientific validity of our forensic reports, we adhere to rigorous international standards. All reports are designed to be subject to peer review and include detailed information such as hash values, allowing for verification of the original evidence. To anticipate potential challenges and scrutiny from legal proceedings, our team includes experts from renowned institutions such as Stanford University, Yale University, Institute of Atomic and Molecular Sciences, Academia Sinica., National Taipei University of Technology, and National Taiwan University of Science and Technology. We have also implemented strict protocols and procedures to prevent any bias or misconduct. Our forensic reports stand apart from traditional expert reports by providing a level of scientific rigor and transparency that is often lacking. Unlike subjective opinions or black-box software results, our reports are structured to meet international standards for scientific publication. This allows for independent verification by peers, ensuring the objectivity and reliability of our findings. Our reports clearly outline our conclusions, as well as any unresolved issues or areas for further investigation. A) The "Findings" section of our reports presents objective, data-driven results from various tests. Our expert team provides a comprehensive analysis of each test, including potential causes, significance, and real-world implications. These findings are presented independently, allowing for thorough scrutiny by peers, judges, prosecutors, investigators, or any third party. Each individual test result can be independently verified and collectively analyzed to form independent conclusions. Our approach ensures a clear and unbiased presentation of scientific evidence. B) While the previous section presents objective data and findings, the "Open Questions, Overall Opinions, and Conclusions" section offers our expert interpretation of these results. Our opinions are informed by our extensive experience and knowledge in the field of forensic audio analysis. We may also identify areas where additional information is needed to draw more definitive conclusions. Our overall conclusions provide a comprehensive assessment of the case, combining both objective data and expert judgment.
Q10. Can AI-generated voice files be authenticated? Can your laboratory detect deepfake and deep voice audio generated by popular software? Since AI voice synthesis is inherently a form of manipulation, our laboratory's expertise in file alteration analysis is well-suited to determining the authenticity of AI-generated audio. By comparing voice prints and analyzing linguistic features, we can effectively verify the genuineness of such files. Moreover, most current AI voice synthesis technologies and simulations are developed using open-source toolkits provided by major international companies, which are then used to train neural networks with large datasets. However, most developers, technology companies, or malicious actors lack expertise in linguistics, behavioral sciences, audio engineering, or psychoacoustics. Leveraging our existing research capabilities, experimental data, and collaborations with major international companies in developing AI voice synthesis and TTS systems, our laboratory is well-positioned to maintain a significant technological advantage in detecting AI-generated audio forgery for the next 10-15 years.
Q11. Guidelines for Fact-Checking Public Complaints and Whistleblowing Allegations Against Public Officials and Media Personnel. In recent years, there has been a surge in the use of manipulated or fabricated evidence by politicians, media personalities, influencers, and pundits to attack their opponents. This has led to a proliferation of misinformation in the news media, making the public vulnerable to disinformation and cognitive warfare. MuScene Voice Forensics Laboratory has frequently been commissioned by elected officials and media professionals to verify the authenticity and originality of audio files. Based on our experience, we offer the following guidelines for those seeking to fact-check online allegations or conduct due diligence before releasing audio evidence: 1. Time-stamp accuracy: Obtain the precise timestamp (year, month, day, hour, minute, second) of the recording. 2. Device and software details: Collect information about the recording equipment and software used, including specific settings. 3. Recording environment: Document the recording location, acoustics, and any ambient noise. 4. Avoid digital alteration: Minimize the risk of file corruption by avoiding digital transmission platforms like Line. 5. Hash value verification: Implement hash value verification to ensure the integrity and authenticity of the audio files. The more details we have about the recording, including the device used, recording conditions, and timestamps, the more accurate our analysis will be in determining the file's authenticity and provenance. For our clients who have entrusted us with audio/video forensics: 1. Share your findings: Even if the initial investigation is incomplete, consider sharing some or all of the results to promote transparency. 2. Acknowledge our laboratory: Clearly state that the investigation was conducted by our laboratory to avoid unnecessary inquiries. 3. Coordinate communication: Inform our contact person if you plan to hold a press conference or release a statement, and authorize the scope of information we can disclose. 4. Accurate representation: Please avoid exaggerating our findings or misinterpreting scientific data. This can lead to the spread of misinformation and negatively impact public opinion. 5. Confidentiality and integrity: While we maintain strict confidentiality, we also uphold principles of transparency. Misrepresenting our findings or using our name to endorse false information is strictly prohibited.
