Introduction
Correct the right problem before changing the name or number
Start by distinguishing what the narrator said from who or what the statement identifies. A transcript may misspell a clearly identified relative, faithfully record a nickname, or contain a name whose referent remains uncertain. Those are different problems. Listening can help establish the spoken words in an actual recording. A narrator's clarification may establish a spelling or explain a relationship. A historical document may offer evidence about a person, but it does not automatically establish that the person is the one mentioned in the interview. Review the particular question before replacing the text. Otherwise a plausible discovery can cause an accurate transcript to be rewritten around the wrong relative. The useful goal is a readable account that preserves the distinction between the narrator's words and the research used to explain them, with enough source context to revisit a disputed decision later.
The authored family example follows a fictional narrator, Nora, who describes two relatives named Elsie Marsh. One is her aunt, called Essie; the other is that aunt's cousin. Nora gives different spouses, discusses an uncertain envelope spelling, corrects a moving year, and recalls an address and household count with qualifications. Three fictional record descriptions supply possible comparison material. None comes from a real archive, family, photograph, or genealogy profile. The example is deliberately constructed so that a familiar name and a roughly compatible age cannot resolve every identity question. It shows how to organize a review, not how to identify an actual ancestor from a recording that was never supplied. In a real project, the audio, narrator, and relevant records must provide the evidence that the authored example can only illustrate.
References: National Park Service transcription guidance; FamilySearch guidance on repeated names in families.
Keep the interview account and the research conclusion connected but distinct
Separate the transcript, the clarification note, and the identity conclusion. The transcript represents the interview under the project's editing conventions. A clarification can explain a spelling, nickname, or reference and identify where that information came from. An identity conclusion connects the mention to a particular person using the available evidence. Keeping these roles visible prevents an editor's inference from becoming something the narrator appears to have said. It also makes correction easier: a new record may change the identity assessment without changing the original spoken nickname. National Park Service guidance includes narrator review and explanatory notes as part of transcript preparation. That supports an accountable review process, not a reason to conceal later additions inside an apparently untouched record of speech. Decide what each change does and preserve the relationship that lets another reader understand it.
The review packet can remain modest. Keep the source recording, the current transcript, and a short record of material name or number decisions. For the exercise, stable references F01 through F10 identify the authored dialogue, while D1 through D3 identify the fictional record descriptions. They are text identifiers, not invented audio timestamps or archival references. A real packet should use locators that lead to the actual source. If a family member supplies a correction after the interview, identify that contribution rather than making it indistinguishable from the recorded account. If an external record suggests a different person, retain the reason for that assessment. This approach preserves useful research without asking the transcript to perform every historical task at once. The account remains available even while the identity or chronology question continues to be investigated.
References: National Archives guidance on citing records; Library of Congress personal audio preservation.
1. Identify the uncertainty in the actual passage
1.1 Separate hearing, spelling, identity, and relationship questions
Keep the audible form, the written form, the person mentioned, the relationship, the event, the evidence source, and the remaining uncertainty separate. In F01, Nora explicitly describes two Elsie Marshes. A later record containing that name cannot be assigned to one relative merely because its spelling matches. In F02, she says her aunt was called Essie and that she never called her Elizabeth. Expanding Essie to Elizabeth would contradict the supplied account rather than improve it. These are authored text observations, not listening results. In a real recording, first establish the words as far as the audio supports them, then investigate spelling and identity using the relevant context. A correction that answers one question may leave another unresolved. Clear spelling does not guarantee a unique person, and a strong identity lead does not necessarily reveal how the narrator pronounced a name.
Preserve the relationship wording when it helps distinguish people. Aunt, cousin, neighbor, and a person's name can play different roles in a family account. Nora's aunt and the aunt's cousin should not be merged because they share a surname and first name. At the same time, an informal family term may need clarification before it is translated into a formal relationship in a research record. Ask what the narrator means where the source leaves that uncertain. Do not silently impose a standard pedigree relationship simply because a database requests one. The transcript can preserve the original expression while a note explains what has and has not been established. That gives a later researcher room to compare records without inheriting an editorial assumption disguised as a statement from the interview. Even the direction of a relationship deserves attention. My aunt and her cousin identifies the cousin in relation to the aunt; shortening it to my aunt and my cousin would change the supplied statement. A family reader might recognize the intended people, but the transcript should not rely on that private familiarity to repair a changed relationship. Keep the referent clear when the passage is summarized or annotated. If the source itself leaves the relationship ambiguous, ask for clarification rather than choosing the branch that makes a preferred record match easier.
References: FamilySearch guidance on repeated names in families; FamilySearch guidance on evidence before merging.
1.2 Treat nickname expansion and surname normalization as decisions
Distinguish a nickname, a documented name, a spelling variant, a proposed equivalence, and a confirmed connection. Familiarity can make an expansion seem obvious, especially when a family tree already contains a likely person. The exercise deliberately resists that shortcut. Essie is the supplied nickname, Elsie Marsh is the name Nora gives, and Elizabeth is a name she explicitly rejects for her own usage. Retain those distinctions. If a real source later documents a formal name, add the supported information in the appropriate explanatory role rather than replacing every instance of the nickname in the transcript. A readable account can explain that two forms refer to the same person when the evidence supports it. It should not manufacture the connection merely to make the names conform to a preferred spelling or an existing profile.
The surname issue in F04 is different. Nora says the old envelope looks like March, although she remembers Marsh. The exercise includes no envelope image, so there is no handwriting result to report. The statement is Nora's description of a possible discrepancy. A reviewer can preserve that discrepancy and identify what would need inspection. An external index using Marsh might be relevant, but it cannot settle the envelope's handwriting without the actual document. Likewise, the envelope's form would not by itself establish which Elsie was intended. Keep the scope of each source visible. A spoken memory, a visual reading, and a person identification are related inquiries, but treating them as one correction can remove the very uncertainty that a later reviewer needs to understand why several candidate names remain under consideration.
References: National Park Service transcription guidance; FamilySearch guidance on evidence before merging.
1.3 Read a number with its event, unit, and qualification
Retain the number, its unit, the event it describes, its precision, and the speaker's qualification. In F05, Nora says the family moved in 1958, then corrects that to 1959. A reviewed account can make the self-correction clear under its editing conventions. That establishes what the authored statement says, not that external historical evidence proves the move occurred in 1959. In F06, she says someone was about twenty when they moved but is unsure of the year. Do not combine the two turns automatically into a precise birth-year calculation. The pronoun and the event relationship still need attention: we moved and they moved are not necessarily the same event. Arithmetic can be flawless while the connection between its inputs is unsupported. Establish the relationship before deriving another historical value from the numbers.
Preserve approximation and correction in a way the next reader can understand. About twenty is not exactly twenty, and four at the time is not necessarily the total number of children ever in a family. F09 first mentions five children, then says four at that time, with a baby arriving later. Replacing the entire account with four children would remove the event context; retaining only five could misdescribe the earlier household. The useful review asks which population and moment the number concerns. In an actual interview, replay the surrounding exchange when the count affects the research. Do not remove a hesitation, correction, or time phrase simply because the resulting sentence is cleaner. Those small expressions can carry the information needed to distinguish a remembered household from a final family total.
References: National Archives census clues and age limits; National Park Service transcription guidance.
2. Review the wording before adding outside information
2.1 Use the recording and its context for transcription questions
Start with source identity, the relevant passage, speaker attribution, surrounding wording, and the review outcome. For a real family recording, listen to the complete exchange around an uncertain name or number rather than replaying a tiny fragment with one preferred answer in mind. An interviewer may suggest a number that the narrator does not accept. A later sentence may identify which relative was meant. If the audio remains unclear, preserve that result rather than selecting the spelling that fits the existing family tree most conveniently. The transcript should make the uncertainty recoverable. Keep the source location so another reviewer can examine the same passage with its context. A historical record can help formulate a candidate, but it should not be treated as proof that the candidate is the word actually audible in the recording.
With WhisperAITranscribe, an online audio and video transcription tool, the resulting text can be a starting point for that review. This guide does not test recognition of family names, old recordings, handwriting, or historical identity. It does not claim a family-tree connection or an automatic archival verification feature. The example uses authored text references because there is no audio to inspect. In a real workflow, retain the actual recording and the reviewed passage locations rather than fabricating precise time references. The transcription stage and the later research stage should remain connected, but each must supply its own evidence. That makes a correction easier to explain: the reviewer can say whether the words were clarified by listening, a spelling was supplied by the narrator, or a separate source suggested an identity that still needs evaluation.
References: National Park Service transcription guidance; Library of Congress personal audio preservation.
2.2 Ask the narrator for the particular clarification that is missing
Separate the uncertain passage, the clarification question, the supplied answer, its date, and its effect on the transcript. A useful query can identify the exact name or number and explain what remains uncertain. Ask which Elsie was meant, how the surname is spelled, or whether the two moving references concern the same event. Avoid presenting a preferred identity as an established fact inside the question. The aim is to obtain clarification, not invite agreement with an editor's existing theory. National Park Service guidance includes checking proper names and narrator review. In practice, retain the contribution's origin so the next reader can distinguish the recorded wording from information added during review. A narrator's later explanation can be valuable without becoming a sentence that appears to have been spoken in the original interview.
The answer may resolve only part of the problem. A family member might confirm Marsh as the remembered surname but remain unsure which document belongs to which relative. They might explain that Tom was the name used in the household without knowing whether a particular Thomas in a record is the same person. Keep that narrower result. It is progress, even when it does not complete the identity investigation. If the narrator changes a remembered date after consulting another source, identify the later basis for the revision where it matters to the project. The review record should make the change understandable rather than collapse listening, memory, and documentary research into an anonymous corrected label. That preserves the value of the interview and the value of later clarification without pretending that they are identical forms of evidence.
References: National Park Service transcription guidance; National Archives guidance on citing records.
2.3 Keep an interviewer's suggestion separate from the narrator's answer
Track the question, the answer, the speaker, and the remaining uncertainty. In F07, the interviewer asks whether the house number was fourteen. Nora responds in F08 that she remembers forty, while noting that the photograph is unclear. The supplied account does not support silently choosing fourteen because it appears first or because the question sounds confident. It also does not justify claiming that a photograph was examined and found to show forty. No image exists in this exercise. Preserve the actual roles: fourteen is the interviewer's suggestion; forty is Nora's qualified recollection; the photograph remains an uninspected source described as unclear. This is exactly the kind of small exchange that can be flattened when names and numbers are extracted without their speakers. The words remain present, but their evidential meaning changes.
For a real recording, inspect how the narrator responds to a suggested spelling, date, or relationship. Agreement, correction, uncertainty, and a change of topic should not all become confirmation. If a later note summarizes the result, keep the qualification that affects the research. A short entry such as house number recalled as forty, photograph unresolved can be more useful than a bare number that conceals its status. The phrasing should follow the project's conventions and the actual source, not this example mechanically. The important feature is that another reader can tell what was offered by the interviewer and what the narrator supplied. That makes the next investigation specific: inspect the relevant image or seek another address source, rather than treating a suggested number as already established because it appeared in the conversation. Also avoid converting competing readings into a numerical range. Fourteen and forty are two distinct address candidates in this exchange, not an asserted interval from fourteen through forty. A compact note should retain the candidate status and speaker ownership. That makes a later document check more focused: the reviewer is comparing particular possibilities, not searching a range that the narrator never supplied. This is one reason to preserve the exchange before turning it into a structured field, where a convenient dash or a single number can accidentally change the type of information being recorded.
References: National Park Service transcription guidance; FamilySearch guidance on evidence before merging.
3. Compare records without collapsing different relatives
3.1 Use a candidate record as a lead rather than a completed match
Keep the person mention, the candidate record, and the proposed connection identifiable. The exercise's fictional D1 entry describes Elsie Marsh, age forty-three on April 1, 1950, with spouse Thomas. D2 describes another Elsie Marsh, age forty-one on the same reference date, with spouse Jack. F03 says Aunt Essie married Tom and the other Elsie married Jack. Those details create comparison questions. They do not establish that D1 is the aunt merely because Tom and Thomas might be related name forms. Nor does a matching Jack alone prove that D2 is the cousin. Inspect the complete evidence needed to connect the person mention and the record. FamilySearch's merge guidance emphasizes evidence rather than a mere absence of conflicts. That is useful here because an incomplete candidate can look compatible simply by leaving the distinguishing facts unspecified.
Give each candidate its own working identity while the question remains open. A note can retain D1 and D2 as separate fictional leads in the exercise; an actual project should use real source identifiers and person references. Record the supporting similarities, meaningful differences, and missing information. Do not copy one candidate's age into the other person's transcript note to make the account fit. That would use an uncertain connection to rewrite the evidence needed to evaluate the connection. Keep the reasoning open to revision. If a later source establishes a different spouse, residence, or relationship, the reviewer should be able to revisit the candidate without undoing a web of silent substitutions. This is a practical reason to resist premature merging even when a likely match feels familiar to the family or appears repeatedly in a search result. Repetition can come from copying rather than additional evidence. If several working notes repeat the same earlier identity assumption, identify that shared origin before treating their agreement as corroboration. The point is a reasoning check, not a claim about any actual family website in this exercise. Follow the material statement back to the source that supplied it. A candidate supported by one uncertain description should remain distinguishable from a candidate supported by separately examined evidence, even if both currently appear in several convenient summaries. Counting the summaries would answer a different question from evaluating the underlying connection.
References: FamilySearch guidance on evidence before merging; FamilySearch guidance on repeated names in families.
3.2 Use age as a bounded clue rather than an exact birth date
Distinguish a reported age and a derived birth value. For the U.S. census records covered by its 1850–1950 guidance, the National Archives explains that age refers to the official census day and gives an approximate birth clue rather than an exact date. For the fictional D1 entry, simply subtracting forty-three from 1950 produces 1907, but that arithmetic does not establish a precise birth date. We checked four authored dates locally under the explicit assumption of accurate age in completed years on April 1, 1950. April 1, 1906 gives forty-four; April 2, 1906 gives forty-three; April 1, 1907 also gives forty-three; and April 2, 1907 gives forty-two. The two middle dates demonstrate the ambiguity. They are calculated examples, not candidate birthdays found in a census, and they do not validate the accuracy of the fictional record or identify which Elsie it concerns.
Keep the assumptions with any derived value. A historical age can itself be mistaken, approximate, or reported by someone whose knowledge is limited; the National Archives notes that census accuracy depends on the informant's knowledge. An arithmetic result should not conceal those source conditions. Use the age to guide further comparison and record searching, then seek the appropriate evidence for a stronger date claim. Do not replace a narrator's uncertain birthday with a single calculated year and label it corrected. The review may instead retain the age as reported and explain the range of possibilities it suggests under stated assumptions. This leaves the next researcher able to understand why a record is relevant without mistaking a convenient calculation for a documented birth event. The calculation narrows a question; it does not complete the historical investigation by itself.
References: National Archives census clues and age limits; FamilySearch guidance on evidence before merging.
3.3 Distinguish a documentary description from the document itself
Keep the described source, the inspected source, and the interpretation separate. F04 describes an envelope spelling, and the fictional D3 record description supplies E. March, 1959. Neither is an actual envelope image in this exercise. It would therefore be false to report a handwriting inspection or say the envelope conclusively confirms March. Even with a real image, the initial E and a year would not automatically identify which Elsie was meant. Inspect the actual item and the information needed to understand it, including its context and provenance. A catalog entry or a family note can be a useful route to the item while still being a distinct source. Preserve that distinction when adding a transcript annotation so a later reader can tell whether the editor saw the document or relied on somebody else's description of it.
In a WhisperAITranscribe review workflow, the transcript can preserve Nora's account while an annotation explains the documentary question. The transcription tool does not establish handwriting, archival identity, or the relationship between two historical records. If an actual document is later inspected, cite that item and identify what the inspection supports. It might clarify a spelling while leaving person identity unresolved. It might show that an earlier description was incomplete. Keep the result specific rather than turning one useful observation into proof of the entire family connection. This makes the record easier to maintain because a later correction can target the annotation or identity assessment without erasing the interview account. It also avoids giving a typed catalog description more authority than the source it was meant to help the researcher locate.
References: National Archives guidance on citing records; National Park Service transcription guidance.
4. Keep evidence and derived information traceable
4.1 Give each important correction a source and a reason
Retain the original form, the reviewed form, the reason for change, and the supporting source. A spelling correction supplied by the narrator should identify that contribution. A number clarified by replaying the audio should retain the relevant passage location. A proposed identity based on documents should explain the connection and remaining limits. These records need not be elaborate. Their purpose is to prevent a later editor from inheriting a corrected name with no way to understand why it changed. The National Archives citation guidance describes providing enough information to locate the records being cited. Apply that practical goal to the relevant evidence in the project: the next person should be able to find the recording passage, clarification, or document that supports the decision rather than rely on an unexplained note that says verified.
Use a locator that fits the source. An actual recording may have a file identity and a checked time range. An archival item may require a repository and the relevant record or collection identifiers. A later family clarification may need the contributor, date, and retained message or review form under the project's arrangements. Do not copy a formal archival citation structure onto every informal source as though the same fields exist everywhere. The point is recoverability and clear attribution. In the exercise, F and D identifiers do that job only within the authored packet; they are not substitutes for real citations in a public family history. Keep the limits visible when the example is adapted so a reader does not confuse a convenient internal label with a source reference that can actually be retrieved outside the working document.
References: National Archives guidance on citing records; Library of Congress personal audio preservation.
4.2 Preserve event context when numbers appear to conflict
For a WhisperAITranscribe-derived account, retain the event, the people included, the time reference, the unit, the correction, and the unresolved relationship. F09's four children at the time and a baby arriving later can describe different moments without being a contradiction that needs one number erased. F05's corrected moving year and F06's uncertain age may concern different moves. Establish the event connection before comparing the figures. In a real project, a household count, a total number of children, and the number present in a photograph can legitimately differ because they describe different populations. The review should identify the population that each statement actually supplies. Treating every family number as one universal count can create an artificial conflict and then invite an editor to solve it by choosing whichever figure fits the preferred story.
When two figures still conflict after context is checked, preserve the conflict and identify what evidence might resolve it. An interview can contain a mistaken recollection without being a transcription error. A document can contain an error or refer to another person. Do not choose the typed record automatically over the narrator, and do not choose a familiar family memory automatically over the document. Assess the particular evidence and keep the source account available. The next step may be another record, a clarification, or a return to the actual audio. The task is to make the disagreement understandable, not to remove it at any cost. A well-kept unresolved note is useful because it tells the next researcher exactly which people, event, and figures need comparison rather than hiding the problem behind a single unsupported replacement.
4.3 Keep firsthand recollection and a story heard from someone else distinct
In a WhisperAITranscribe transcript, preserve the narrator, the reported speaker, the event described, and the basis of knowledge. F10 says the mill story came from Nora's father and that she did not see it herself. A summary should not turn that into Nora witnessed the event at the mill. The story can remain valuable as a family account while its attribution stays clear. This distinction also matters for names and numbers: a person may be repeating a date heard from a relative rather than recalling it directly. Retain that information when it affects interpretation. The transcript's job is to represent the account faithfully, including how the narrator says they know it. Historical evaluation can then consider that source relationship instead of starting from a more direct eyewitness claim than the interview ever made.
If a later writer condenses the passage, inspect the attribution after shortening. A sentence that begins Nora recalled can become ambiguous about whether she recalled the event or recalled being told the story. Use wording that preserves the actual relationship. The same care applies when a family member supplies a correction based on another person's recollection. Identify the contribution and its source where it matters rather than collapsing several voices into one anonymous family fact. This is not a demand that every remembered detail carry a formal research essay. It is a practical way to keep a material statement's origin understandable. When the name, date, or identity is disputed, knowing whether the speaker witnessed the event, heard it from a relative, or read it in a document can shape the next useful question.
References: National Park Service transcription guidance; National Archives guidance on citing records.
5. Maintain a readable account without losing its evidence
5.1 Keep later clarification identifiable when the transcript changes
Separate a transcription correction, a later clarification, and a research update. A corrected misheard number concerns what the recording says. A narrator's later explanation may add information that was not supplied in the interview. A new document may change which person an editor considers the best match. These are different revisions. Keep enough context to show which occurred, especially when a name or date affects other parts of the family account. National Park Service guidance includes review and explanatory notes, while the project's own conventions determine how those changes appear in the final transcript. Follow the chosen approach consistently. Do not call an addition verbatim simply because it came from the same narrator at a later time. The information may be useful and authoritative for its purpose while still being a separate contribution from the original recording.
When a correction affects several copies, identify the relevant current version and the places where the old form remains in use. A research note, a reading transcript, and an index may each need attention, but they should not all be overwritten without considering their role. A historical draft can remain evidence of earlier work while the current reading copy is clearly identified. Keep the original recording available so a future reviewer can revisit the question rather than rely only on the latest polished text. The Library of Congress's personal audio guidance emphasizes organizing recordings with descriptive information and managing copies. Use that preservation work to support the review rather than treat the recording as disposable once a transcript exists. The audio and its descriptive context remain useful when later family research asks a question the first editor did not anticipate.
References: National Park Service transcription guidance; Library of Congress personal audio preservation.
5.2 Make uncertainty readable rather than removing it for convenience
Keep the important qualification and the explanation's origin visible in the delivered account. A reader does not need to see every tentative spelling an editor considered, but they should not be led to believe an unresolved identity is settled. A concise annotation can identify which Elsie remains uncertain, which date was self-corrected, or which address number needs a document check. Put the qualification close enough to the relevant claim that it survives ordinary reading. If a short summary or excerpt drops the condition, the resulting account can become more certain than the source even when every retained word is technically correct. Review the delivered form, not only the detailed working notes. The useful question is whether a person who did not perform the research can still understand what is known and what remains open. Use an uncertainty label that identifies the unresolved question. Name unclear can imply that the spoken form is uncertain, while identity unresolved can indicate that the words are clear but the person has not been connected confidently to a record. Those situations call for different next steps. In the exercise, two Elsies are explicitly named; the review concerns keeping their references and candidate records separate. A vague label could conceal that useful information. The delivered account should make the next investigation easier by explaining what still needs evidence rather than simply announcing that the passage is imperfect.
For a WhisperAITranscribe-derived reading copy, preserve the source relationship as names and numbers are clarified. A note can identify a narrator-supplied spelling without pretending the tool verified a historical identity. An external record can be cited as a candidate without being folded into the transcript as an established match. When the account is translated or adapted for another family reader, keep those roles intact. The exercise does not test translation or multilingual recognition, so it supports no accuracy claim about either. The practical editing principle is to preserve the degree of certainty while making the account easier to follow. That lets readers benefit from the work without inheriting a false confidence created by the formatting or by a summary that quietly removed the unresolved part of the claim.
References: National Park Service transcription guidance; FamilySearch guidance on evidence before merging.
5.3 Treat a family-tree update as a separate research action
For WhisperAITranscribe material, keep the person mention, the evidence comparison, the identity conclusion, and the proposed update distinct. A WhisperAITranscribe transcript can preserve the account of two Elsies while the family research continues to evaluate their records. It does not itself authorize or verify a merge. FamilySearch guidance describes the problems caused by incorrect merges and recommends examining evidence before connecting profiles. The broader lesson applies to the research decision: matching names or a lack of obvious conflict is not enough to establish that two incomplete descriptions represent one person. In this exercise, the aunt and cousin remain separate mentions, and the fictional record descriptions remain candidates. No actual profile was inspected or changed. A real update needs the relevant evidence and the appropriate review for the system being used.
If a later conclusion connects a record to a person mentioned in a WhisperAITranscribe transcript, retain the reason and the source references that support it. If the evidence remains insufficient, leave the proposed link unresolved. That outcome is more useful than an unsupported connection that spreads into subsequent notes and family accounts. Keep the original wording available even after the research conclusion becomes stronger, because future questions may concern how the narrator described the person or event. The stopping point for transcript review is not a perfectly completed family tree. It is an account whose important words and numbers have been reviewed, whose additions are attributable, and whose unresolved identities remain visible. Historical research can then proceed from a dependable record without being forced to accept conclusions that the transcription stage was never equipped to establish.
6. Conclusion
6.1 Preserve the person distinctions while improving the transcript
A dependable family-history review connects source wording, name forms, person distinctions, event context, numerical limits, and attributable clarification. The authored example shows why these checks cannot be reduced to choosing the most familiar spelling. Two Elsies remain two people in the narrator's account. Essie need not become Elizabeth. A corrected moving year does not settle an unrelated age reference. A suggested house number is not the narrator's confirmed answer. A census-style age can narrow a birth question without supplying an exact date or proving identity. With WhisperAITranscribe as the transcription starting point, keep those distinctions available throughout the review. The objective is a clearer account of what the source supports, with later research connected to it rather than substituted for the narrator's words.
The useful next action follows from the particular uncertainty. Replay an unclear passage, ask for a spelling, inspect an actual document, compare separate person candidates, or retain a conflict that needs further evidence. Do not let a convenient database field decide the answer before the source does. A transcript can be readable and useful while an identity question remains unresolved. In fact, keeping that question visible can prevent later work from building on the wrong relative. The four date calculations reinforce the same point at a smaller scale: a correct operation can show several possibilities without choosing the historical fact. Keep that boundary when moving from a spoken number to a derived value or from a matching name to a person connection. The review should expose what the evidence establishes and what it still leaves to investigate.
6.2 Follow the source through correction, clarification, and research
Step 1: Preserve the recording and identify the exact passage, speaker, name or number, and surrounding event context. Step 2: In the WhisperAITranscribe workflow, review the wording and obtain any supported narrator clarification without disguising later additions as original speech. Step 3: Compare documentary clues and person candidates separately, keeping assumptions visible when deriving dates or interpreting household counts. Step 4: Save a concise name-and-number review record that connects the current form to its evidence and leaves unresolved identity questions open. The sequence should make each correction explainable. It should not force every uncertain name into an existing profile or turn every age into a full date simply because a downstream form has a field waiting to be filled.
Use the amount of documentation the question needs. A straightforward spelling supplied by the narrator may need only a short note. A disputed identity involving two relatives with the same name deserves a clearer comparison and source trail. Preserve the source recording and the current transcript so the next reviewer can continue the work without reconstructing which version was used. The exercise's ten turns and fictional record descriptions make the distinctions easy to inspect, but they do not replace the evidence required for an actual family. A real project needs its own listening, clarification, and record review. The method is useful because it keeps those tasks separate enough to evaluate while connecting them closely enough that a material name or number can be followed from the original account to the current explanation.
6.3 Leave a clearer account rather than a prematurely completed identity
Use WhisperAITranscribe with family-history review that preserves meaning, attribution, precision, and remaining uncertainty. When a WhisperAITranscribe transcript contains a familiar nickname, a corrected year, or a possible record match, resolve the particular question the evidence can answer. Keep two people separate when identity is unresolved, distinguish a narrator's recollection from an inspected document, and retain the event context that makes a number meaningful. A careful correction improves access to the account without making it more definite than the source permits. That leaves the family with a useful transcript and a research trail that can be continued, corrected, or strengthened later, instead of a polished story whose hidden substitutions have already merged different relatives into one apparently certain person.
References: FamilySearch guidance on evidence before merging; National Archives guidance on citing records.