In Unheard: Voices of Crime, listen to one conversation as a whole before trying to connect every speaker. Jumping between rooms whenever a voice catches your attention can leave you with several memorable lines and no reliable account of who said what to whom.
The developer describes a crime-solving game built around audio: you follow conversations, connect names with voices, and replay recordings while investigating overlapping events. You can follow one person's story or move between people. The recordings can be replayed as often as you need, so missing a line on your first pass is a reason to rewind, not abandon the lead. Official Steam description
First pass: understand the conversation
On your first pass through an unfamiliar exchange, aim to understand the subject and the participants. Let the conversation provide context before deciding which sentence is the crucial clue. An interruption, reply or correction can change the meaning of something said a few seconds earlier.
Use temporary speaker labels when you do not yet know the names. Keep those labels neutral: “speaker A” or a brief voice description is safer than “the killer.” A theory-based label makes it easier to mistake suspicion for identification.
If two voices are difficult to distinguish, replay a short exchange in which the turn-taking is clear. Pay attention to who answers whom. Do not assume that accent, age impression or tone by itself proves an identity or motive.
Second pass: track one person
Once the exchange makes sense, follow one participant through the material available to you. Notice where the person goes, who they meet and whether they use a name for someone else. Your goal is to build a sequence that you can check against other conversations.
Record the point in the recording, the location, and the exchange you want to compare. A short reference such as “after A enters the room” can help you find a passage again without transcribing it.
Do not copy long conversations. A note such as “A is addressed by name during the later conversation” is more useful than half a page of dialogue when your unresolved question is simply who A is.
Third pass: check what happened elsewhere
After following one thread, return to a relevant moment and examine another conversation. This is where an apparently complete account may acquire a missing cause, contradiction or context.
The developer says stories overlap in the same space and time, and audio can be replayed. That does not mean every interesting event explains every other event. Compare the timing and participants before linking two exchanges merely because their subjects sound similar. Developer's explanation of the listening structure
Investigate a precise overlap: “What was happening in the other location while this person made that claim?” That is easier to answer than listening to the whole case again without a question.
Distinguish a statement from a fact
Hearing someone say something proves that the statement was made. It does not automatically prove that its contents are true. Keep three different kinds of note separate:
| Kind of note | Fictional example | What remains uncertain |
|---|---|---|
| Direct observation | A voice calls another person a name | Whether the name is genuine |
| Reported event | Someone says a parcel arrived | Whether the arrival occurred as described |
| Interpretation | You think two speakers are coordinating | Whether another conversation supports that theory |
These are invented practice examples, not case clues. The distinction is especially helpful when a confident voice makes a weak claim sound settled.
Before accepting a major conclusion, ask what independent exchange supports it. Repeating one person's claim in a different note does not turn it into a second piece of evidence.
Look for another conversation before looking for an item
NEXT Studios explicitly describes cases with no hidden information or rooms requiring keys. The evidence is in the audio. If a question remains unresolved, follow another participant through the relevant time period rather than searching for an inventory item that might unlock the next clue. Developer's description of the case structure
Fix a confused identity without restarting everything
When a name assignment begins to contradict the recordings, return to the first reason you made it. Was the person directly addressed, or did you infer the name because only two possibilities seemed available?
Correct the earliest uncertain link and then review the conclusions that depended on it. This contains the repair. Replaying everything from the beginning may be unnecessary when one mistaken speaker assignment created the confusion.
Keep confirmed and provisional entries visibly different in your own notes. A question mark is enough. Without that distinction, your later self may treat a hurried guess as something the game explicitly established.
Choose a listening setup you can sustain
Use a comfortable listening volume and a setup in which dialogue is clear. Headphones are part of the developer's suggested experience, but clarity matters more than turning the volume up. Pause when you cannot give the conversation your attention.
Check the current language and accessibility information for your edition before buying, particularly if you need text to substitute for spoken clues. Do not assume that an interface translation means every recording has matching subtitles. As of 29 October 2026, Steam lists English full audio and interface support, but its subtitle column is blank. That listing does not establish a text substitute for the spoken clues. Current language information
At the end of a session, leave one unfinished listening question and its location. Returning to a specific conversation is much easier than trying to reconstruct a whole cast from memory.
Comments