开云体育

  1. N1MMLoggerPlus
  2. Topics

Super Check Partial Data WAV Files

 

Does anyone know of a way to create WAV FIles for the Super Check Partial Database?
Mark Griffin, KB3Z
 

I started going down that path. Over the weekend I developed code to parse the SCP database, dedupe repeating calls, and then split each call in a pair of 1/prefix and 2/suffix. The logic is that voicing will sound better if the prefix and suffix are spoken separately.
?
I also started experimenting with creating a custom voice using Eleven Labs. My thinking is that 1/it will be very confusing if everyone sounds the same and 2/custom voices can be designed specifically for contesting. Here’s a sample prompt:
?
# Ideal Male Contest Voice (Eastern European)
Create male voice, 55yo with very subtle Eastern European accent:
- Pitch: Medium-low (110Hz), narrow range
- Resonance: Deep chest (60%)
- Accent features: Slight final consonant devoicing, subtle /v/ for /w/, occasional /t/ for "th"
- Speaking style: Fast, authoritative, precise, confident?
- Technical: Enhance 1.8-3kHz range, boost consonant clarity
- Contest optimizations:
? - Compress dynamic range (70%)
? - Maximize intelligibility at speed
? - Clear number pronunciation
? - Subtle micropauses between callsign elements
? - Reduce sibilance (40%)
?
Writing the code to generate the actual voice fragments will follow. I also wonder if I should just record all 50,000 calls in SCP and not bother with recording fragments.
?
Rudy N2WQ
 

开云体育

Gene was gonna use something that they use when making the deep fakes.

Where you read a paragraph to the machine,

then it will make all the calls actually in your own voice.

joe WB9SBD

On 3/30/2025 8:56 PM, N2WQ via groups.io wrote:

I started going down that path. Over the weekend I developed code to parse the SCP database, dedupe repeating calls, and then split each call in a pair of 1/prefix and 2/suffix. The logic is that voicing will sound better if the prefix and suffix are spoken separately.
?
I also started experimenting with creating a custom voice using Eleven Labs. My thinking is that 1/it will be very confusing if everyone sounds the same and 2/custom voices can be designed specifically for contesting. Here’s a sample prompt:
?
# Ideal Male Contest Voice (Eastern European)
Create male voice, 55yo with very subtle Eastern European accent:
- Pitch: Medium-low (110Hz), narrow range
- Resonance: Deep chest (60%)
- Accent features: Slight final consonant devoicing, subtle /v/ for /w/, occasional /t/ for "th"
- Speaking style: Fast, authoritative, precise, confident?
- Technical: Enhance 1.8-3kHz range, boost consonant clarity
- Contest optimizations:
? - Compress dynamic range (70%)
? - Maximize intelligibility at speed
? - Clear number pronunciation
? - Subtle micropauses between callsign elements
? - Reduce sibilance (40%)
?
Writing the code to generate the actual voice fragments will follow. I also wonder if I should just record all 50,000 calls in SCP and not bother with recording fragments.
?
Rudy N2WQ
 

Voice cloning is certainly a viable option. It does work well. On the flip side, it works well if you have a voice that’s good for contesting, otherwise you end up spending time post-processing.
?
Rudy N2WQ

On Mon, Mar 31, 2025 at 02:11 AM, Joe WB9SBD wrote:
Gene was gonna use something that they use when making the deep fakes.

Where you read a paragraph to the machine,

then it will make all the calls actually in your own voice.

joe WB9SBD

On 3/30/2025 8:56 PM, N2WQ via groups.io wrote:
I started going down that path. Over the weekend I developed code to parse the SCP database, dedupe repeating calls, and then split each call in a pair of 1/prefix and 2/suffix. The logic is that voicing will sound better if the prefix and suffix are spoken separately.
?
I also started experimenting with creating a custom voice using Eleven Labs. My thinking is that 1/it will be very confusing if everyone sounds the same and 2/custom voices can be designed specifically for contesting. Here’s a sample prompt:
?
# Ideal Male Contest Voice (Eastern European)
Create male voice, 55yo with very subtle Eastern European accent:
- Pitch: Medium-low (110Hz), narrow range
- Resonance: Deep chest (60%)
- Accent features: Slight final consonant devoicing, subtle /v/ for /w/, occasional /t/ for "th"
- Speaking style: Fast, authoritative, precise, confident?
- Technical: Enhance 1.8-3kHz range, boost consonant clarity
- Contest optimizations:
? - Compress dynamic range (70%)
? - Maximize intelligibility at speed
? - Clear number pronunciation
? - Subtle micropauses between callsign elements
? - Reduce sibilance (40%)
?
Writing the code to generate the actual voice fragments will follow. I also wonder if I should just record all 50,000 calls in SCP and not bother with recording fragments.
?
Rudy N2WQ
 

I don't know if this is possible or not. But in my limited understanding as long as you have WAV Files for A thru Z and 0 thru 9 you should be able to operate using synthesized AI.
But how you get that to work in the N1MM Function Keys is something I don't understand at all. I'm sure if it was that simple everyone would be doing it.
Or, am I just thinking too simple minded?
Mark Griffin, KB3Z
 

Mark, the documentation for N1MM support of WAV callsign and serial voicing is at:?

Basically in your SSB macro you use "!" for callsigns and "#" for numbers and it works incredibly similarly to those macros in CW macros.

There were at least two stations on for WPX this weekend using the K3CT-supplied A-Z, 0-9, etc files you can download at:?

A techie note on WAV files: Almost all Windows media-player programs randomly truncate a fraction of a second from the beginning and end of WAV files,?so previewing a set of voicing macros in WMP or other windows audio player will confuse you. N1MM to its credit does it right and plays the entirety?of the?WAV file.

There were at least 4 stations (including mine, Tina@N3QE) using more advanced synthesis techniques in ARRL DX SSB and WPX SSB this month.

Tim N3QE

On Mon, Mar 31, 2025 at 9:25?AM KB3Z via <pastormg2=[email protected]> wrote:
I don't know if this is possible or not. But in my limited understanding as long as you have WAV Files for A thru Z and 0 thru 9 you should be able to operate using synthesized AI.
But how you get that to work in the N1MM Function Keys is something I don't understand at all. I'm sure if it was that simple everyone would be doing it.
Or, am I just thinking too simple minded?
Mark Griffin, KB3Z
 

Tim,
Okay, I understand that I download the Letter and Number files that John created.
But the confusion that I have is what should the Function Keys Definitions look like for Run in N1MM. I do not see any pictures or anything explaining what needs to be where and what it is going to do when you hit the F1, F2 etc, etc.
Mark KB3Z
 

Mark,
?
It's the same as any other key file, along with Tim's notes on where to put ! for calls and # for numbers.? Everything needed is in the N1MM docs.? Do you have a specific question?
?
73, Drew K3PA
 

For the CQ WPX exchange using voicing of callsigns and serial numbers, for Run mode this could involve basically two changes from the function key messages without using voicing, namely the presence of the # macro in F2 for the serial number and the ! macro in F5 for the other station's callsign:
?
F1 CQ,{OPERATOR}\CQ.wav
F2 Exch,{OPERATOR}\FiveNine.wav,#
F3 TNX,{OPERATOR}\Thanks.wav
F4 {MYCALL},{OPERATOR}\MyCall.wav
F5 His Call,!
?
Similarly for the S&P exchange function key message (F2), and for any function keys you set up for sending repeat serial numbers, etc.
?
73,
Rich VE3KI
?
?
On Mon, Mar 31, 2025 at 10:41 AM, KB3Z wrote:

Tim,
Okay, I understand that I download the Letter and Number files that John created.
But the confusion that I have is what should the Function Keys Definitions look like for Run in N1MM. I do not see any pictures or anything explaining what needs to be where and what it is going to do when you hit the F1, F2 etc, etc.
Mark KB3Z
 

Drew, Rich,
I think I have what I need to do to get this working. First off, I need to download and put the K3CT Letter and Number files in the correct directory, wherever that might be within N1MM.
Next I have to create the CQ.WAV, FiveNine.WAV, Thanks.WAV and MyCall.WAV files and put them in the correct directory within N1MM.
Finally I need to make the necessary changes to F1 thru F5 Function Key designations to get this to work.
Am I correct in what I am saying?
Mark Griffin, KB3Z
 

On Mon, Mar 31, 2025 at 11:22 AM, KB3Z wrote:
I think I have what I need to do to get this working. First off, I need to download and put the K3CT Letter and Number files in the correct directory, wherever that might be within N1MM.
Next I have to create the CQ.WAV, FiveNine.WAV, Thanks.WAV and MyCall.WAV files and put them in the correct directory within N1MM.
Finally I need to make the necessary changes to F1 thru F5 Function Key designations to get this to work.
Am I correct in what I am saying?
Mark Griffin, KB3Z
 

 

Are you answering your own question, or did you have a question?
?
You described the basic idea below, just do it and ask questions if you can't make it work.? It's all well documented, just follow the instructions.
?
73, Drew K3PA
 

I do not have a Exchange.wav or Thanks.wav file created.
So I assume I must go into audacity and create them.?
What do I say when it comes to the Exchange.wav file? Is it simply fve niine and leave it at that?
I assume the Thanks is just simply thanks.
Any help here?
Mark KB3Z
 

 

 

This is the error message that I get when I hit F1. And here is a copy of what my Function Keys have. I am totally confused.
Marrk KB3Z
 

A file name cannot have a space in it...


Ron, WV4P


On Mon, Mar 31, 2025 at 6:40?PM KB3Z via <pastormg2=[email protected]> wrote:
This is the error message that I get when I hit F1. And here is a copy of what my Function Keys have. I am totally confused.
Marrk KB3Z
 

Hopefully I understand this now. The filename should be CQContest.wav. But how should it be listed in the Function Key Definition?
Mark KB3Z
 

For obvious reasons, it has to be the same as the file name, computers are not good at guessing.

Ron, WV4P?

On Mon, Mar 31, 2025 at 7:01?PM KB3Z via <pastormg2=[email protected]> wrote:
Hopefully I understand this now. The filename should be CQContest.wav. But how should it be listed in the Function Key Definition?
Mark KB3Z

1 - 20 of 46