How to Hear Voice Comms in CS2 and FACEIT Demos

Reviewing a demo and hearing nothing but footsteps? Voice in Counter-Strike 2 demos is off by default, and the command that turns it on takes a number most people never work out. Worse, the command is often not the problem at all - plenty of demos have no voice in them to begin with.

Start there, because it decides whether anything else on this page can help you.

Just want the commands? Upload your demo to the CS2 Voice Comms Calculator , tick the players you want to hear, and copy the ready-made tv_listen_voice_indices commands. The rest of this page explains why they work.

Which CS2 demos actually contain voice comms?

Valve matchmaking and Premier demos contain no voice audio data. Not muted, not encrypted - absent. The tool that extracts voice tracks out of Counter-Strike demos carries the warning at the top of its README: there is nothing in an MM demo to extract.¹ No console command can play back audio that was never written to the file.

The second half is the part most guides skip. Voice on a SourceTV server is not automatic either. In CS2, SourceTV does not capture player voice unless the server turns it on, which historically meant sv_alltalk 1 or a plugin doing it for you.² FACEIT records comms and tells players how to play them back.⁵ A pickup server someone rented last week very possibly does not.

Demo source	Contains voice?
FACEIT	Yes
ESEA and other leagues	Yes
Community SourceTV server	Only if that server enabled it
Valve matchmaking	No
Valve Premier	No

So if a demo is silent, the question is not "what did I type wrong" until you have confirmed the demo could ever have had voice in it.

What does tv_listen_voice_indices do?

It decides whose voice you hear while watching a demo through SourceTV. Its default is 0, and 0 means nobody.²

The value is a bitfield of player slots:

• -1 - hear everyone, because -1 is every bit set
• 0 - hear nobody, the default and the reset
• any other number - hear the specific set of slots whose bits are set

The _h command is not a copy, it is the other half

This is the single most misunderstood part, and most guides repeat the mistake.

tv_listen_voice_indices and tv_listen_voice_indices_h are the two halves of one 64-bit bitfield. The first holds the low 32 bits, addressing player slots 0 to 31. The _h one holds the high 32 bits, addressing slots 32 to 63.² CS Demo Manager generates them exactly that way: one loop, one bit per player id, low value for ids under 32 and high value for ids from 32 up.³

Which changes the advice you will read everywhere:

• To hear everyone, set both to -1. That is the recipe FACEIT publishes,⁵ and it works because -1 fills every bit of both halves.
• To hear specific players, put the mask in tv_listen_voice_indices and set _h to 0. In a 5v5 nobody is in slot 32 or above, so _h has no one to address. Copying the same number into it does nothing useful, and on a big SourceTV server it would unmute a mirror-image set of high slots you did not choose.

Setting _h to 0 also clears whatever it was left at earlier in the session, which is worth doing on its own.

How the slot bitmask works

Each bit maps to one slot. Slot 1 is bit 0, slot 2 is bit 1, slot 3 is bit 2, and so on - each slot is worth double the one before. Add up the values of the slots you want:

Slot	Bit value
1	1
2	2
3	4
4	8
5	16

Slots 1, 3 and 4 give 1 + 4 + 8 = 13:

tv_listen_voice_indices 13
tv_listen_voice_indices_h 0

The voice comms calculator does the addition for you. If you would rather work it out from a binary pattern, our number base converter turns it into the decimal CS2 wants.

Finding out who is in which slot

Slot order comes from the player entity order recorded in the demo, so it is not the scoreboard order and there is no reason it should match your intuition. It is, however, fixed for that demo - the same file replays with the same slots every time. What does not carry over is the mask itself: a number that isolated your IGL in Tuesday's match will point at a stranger in Thursday's.

The manual way is to step through them:

spec_player 1
spec_player 2
spec_player 3

and note who appears. The faster way is to upload the demo to the calculator, which reads the slot table out of the file and labels each one with the player's name and team, so you skip the loop.

Hearing one team

Once you know which slots belong to which side, add that side's values together. The calculator has All / All T / All CT buttons that build the number for a whole side in a click, which is what you want for casting or cutting highlights, where ten overlapping voices bury the round.

One caveat worth knowing if you use CS Demo Manager's voice_ct and voice_t aliases instead: they refer to each player's starting side, not their current one.⁴ After the half they no longer mean what the names suggest. The same tool also exposes voice_STEAMID64 for a single player.⁴

If the side you want is the other one, hearing the enemy team is where that gets complicated, because whether their voice is in the file at all was decided by the server during the match.

Reading text chat as well

Voice is half the record. To bring in the in-game text chat:

tv_relaytextchat 2

2 relays everything including all-chat, and 0 hides chat entirely. FACEIT includes it in the three commands it hands out for demo review.⁵ It earns its place when you can hear a call but cannot tell whose voice it is.

They reset every time you close the game

Both convars go back to 0 when CS2 exits. This is the "it worked yesterday" trap: nothing broke, the session simply started from the default of silence again. Keep the commands somewhere you can paste from rather than rebuilding the number each time.

Putting it together

1. Confirm the demo can contain voice - FACEIT or a league, not MM or Premier.
2. Find which players sit in which slots, or let the calculator read them.
3. Add up the bit values for the players you want.
4. Set tv_listen_voice_indices to that number and tv_listen_voice_indices_h to 0. For everyone, set both to -1.
5. Add tv_relaytextchat 2 if you want chat, then play the demo.

Skip the arithmetic and the spec_player loop with the CS2 Voice Comms Calculator : upload the demo, tick the players, copy the commands.