/
2. Clipping process (Research Assistants)

2. Clipping process (Research Assistants)

Introduction to clipping audios, file naming and usual order of administration

 

This section provides guidance for Research Assistants on clipping Connected Speech task audios, including file naming and the standard order of task administration. The same task and clipping order applies across all languages unless otherwise noted by the clinician in the last comment column of the Smartsheet.

Before April 2024 and participant BISE013 all participants were video clipped. Now we only clip audios and in specific cases we clip videos, please refer to video clipping guide: https://cloud.wikis.utexas.edu/wiki/spaces/MADRWiki/pages/108200558

To clip m4a zoom files, please download the FFmpeg before clipping a m4a file audio so it works, Download the correct FFmpeg version for your system: Audacity FFmpeg installer page: https://support.audacityteam.org/basics/installing-ffmpeg After installing, restart Audacity.

Clipping and reclipping should always be completed in the lab. If you must work remotely, it’s essential to follow these steps since we are handling patient data:

  • Connect through the UT VPN

  • Work in a private, secure location

  • Use headphones to protect audio confidentiality

  • Delete all audios and videos from the computers and Trash before finishing your shift

The usual task orders are organized by their sets. Use the linked documents from the REDCap to see clinician prompts for each set.

Important:
Per Decision 20241003, always prioritize clipping Pre, Post, and Follow-up sessions. Mid-treatment sessions should be clipped only if specified for a research need or if all other samples are clipped. This structure ensures consistency, clarity, and efficient data handling across all projects.

Clipping process (important! read before starting to clip)

  • Wear headphones when clipping audios (e.g. Bluetooth headphones, wired headphones)

  • Start clipping exactly when the participant begins speaking.

  • Stop immediately after the participant finishes.

  • Do not include extra silence before or after, as this affects acoustic measures (e.g., words per minute, pause duration).

    1. Examples of comments to omit at beginning and end:

      1. this is hard./esto es difícil/això es dificil

      2. that's all I remember./esto es todo lo que recuerdo/això es tot el que recordo

      3. Done./hecho/fet

      4. that’s all, all done/ ya esta/ja esta

Samples have to have at least two utterances (sentences) if there’s less, clinician will ask (quieres añadir algo más? vols afegir alguna cosa més?)

  • Check the “Latest Comment” section for clinician notes on which sample to use (often the second).

  • Tag the primary and secondary clinician in the channel. Their names can be found in the Smartsheet.

Instruction added to the guide 20250911

If clinician speech appears in the sample:

  1. Identify Clinician Speech

    • Listen carefully for the section where the clinician begins speaking.

  2. Clip Out Clinician Section

    • Start clipping at the moment the clinician begins speaking.

    • Stop clipping immediately when the clinician finishes speaking.

    • Remove this section from the participant’s speech sample.

  3. Check for Smoothness

    • Play back the edited file to ensure the cut flows naturally, without long silences or abrupt cuts.

  1. Clip out the clinician’s speech as much as possible.

  2. If the participant and clinician speak simultaneously, keep that portion in the sample.

  3. Record this occurrence in the corresponding row of the sample in Smartsheets

  1. If background noise occurs when participant is speaking, clip out as much as possible without affecting participant’s speech

  2. For each audio sample, fill out the field “TaskName_Background noise.” Use the following definitions:
    Definitions

  • Compression: Distortion caused by recording or transmission issues (e.g., rushed/zoomed audio, choppiness, or Wi-Fi glitches).

  • Buzzing: A steady, continuous background hum or buzz throughout the sample. (e.g., fan noise)

  • Background noise: Environmental sounds unrelated to the task (e.g., dog barking, people talking, door slamming, microphone tap).

    • NOTE: If a patient’s breathing affects their intelligibility, this would also be considered as background noise

    • Clip it out if possible! Previously clippers where instructed to clip it out when possible

      • If you notice that there is a background noise please describe it as a comment in the participant row stating in which task it's found example: “Picnic Scene Task-There is clinician speech in the beginning of the sample before the patient begins to speak, and a squeaking noise while the patient is speaking.”

  1. If you notice a sample missing in Box, refer to the SmartSheet to see who clipped the audio

  2. Tag the student clipper and student lead in the Clipping/Whisper channel on Teams asking them about the audio’s whereabouts

If an audio sounds like it needs to be reclipped, please send a message in the Clipping/Whisper Channel on Teams: making sure to link the sample, tag the clipping student team lead, and describe your concerns with the audio!

Usual order of task administration and clipping

SET​

TASK FOLDER​

AUDIO NAME (CODE001_BACC001_Taskname_language (Spa, Cat, Eng)_Timepoint_date)​

1​

  • S1_CatRescue_PictureStoryDescription ​

CODE001_BACC001_CatRescue_Spa_Pre_20230914.wav​

3​

  • S3_ImportantEvent – QAB Discourse​

CODE001_BACC001_ImportantEvent_Spa_Pre_20230914.wav​

1​

  • S1_Sunday_Personal Narrative​

CODE001_BACC001_Sunday_Spa_Pre_20230914.wav​

3​

  • S3_Brushing teeth _ Procedural Discourse ​

CODE001_BACC001_Teeth_Spa_Pre_20230914.wav​

1​

  • S1_CatRescue_Recall​

CODE001_BACC001_CatRescueRecall_Spa_Pre_20230914.wav​

1​

  • S1_MAINDog_Story Generation​

CODE001_BACC001_MAINDog_Spa_Pre_20230914.wav​

1​

  • S1_Frog, Where are you?_Wordless Picture Book Story Narration​

CODE001_BACC001_FrogWhere_Spa_Pre_20230914.wav​

3

  • S3_PicnicScene – Picture Description

CODE001_BACC001_PicnicScene_Spa_Pre_20230914.wav

1​

  • S1_Bridge_VIDEO connected speech task​

Task administration prompts: 

"Ahora le voy a mostrar un vídeo en el que salen unos personajes. Por favor, mire el vídeo con atención, cuando acabe de verlo le haré unas preguntas sobre lo que ha visto."

Cuando el paciente termine de explicar el video hagale las siguientes preguntas: 

  1. ¿En el vídeo había varios personajes, puedes nombrarlos por favor?

  2. ¿La historia del vídeo ocurrió en la naturaleza, por favor describe con el mayor detalle el entorno, el lugar dónde ocurrió.

  3. ¿En el vídeo, habia un reno y un oso que hicieron varias acciones, por ejemplo el oso estaba empujando, por favor describe con el mayor detalle todas las acciones que hizo el oso.

  4. ¿También habia un reno, por favor describe con el mayor detalle todas las acciones que hizo el reno.

  5. ¿Aunque el reno y el oso no pueden hablar, seguro que pensaron muchas cosas. Qué crees que pensaba el reno cuando vio al oso por primera vez? ...

  6. ¿Qué crees que pensaba cuando vio al mapache al final?

  7. ¿Qué crees que pensaba el oso?

  8. ¿Cuál era la actitud del oso al principio y al final de la historia? Por favor describe las emociones que sintió el oso.

  9. ¿Cuál crees que es la moraleja de esta historia?

CODE001_BACC001_ Bridge_Spa_Pre_20230914.wav​

After 20250929 we decide to stop clipping these task in 9 audios until we have a more concrete decision on linguistic-acoustic analysis and also to control for previous variability in administration that made the clipping a complex process (see issues described in https://cloud.wikis.utexas.edu/wiki/x/8IO9H and administration variability described here https://cloud.wikis.utexas.edu/wiki/x/UgVjF previously audios were named: CODE001_BACC001_ Bridge_Spa_1_Pre_20230914.wav​

CODE001_BACC001_ Bridge_Spa_2_Pre_20230914.wav​

CODE001_BACC001_ Bridge_Spa_3_Pre_20230914.wav​

CODE001_BACC001_ Bridge_Spa_4_Pre_20230914.wav​

CODE001_BACC001_ Bridge_Spa_5_Pre_20230914.wav​

CODE001_BACC001_ Bridge_Spa_6_Pre_20230914.wav​

CODE001_BACC001_ Bridge_Spa_7_Pre_20230914.wav​

CODE001_BACC001_ Bridge_Spa_8_Pre_20230914.wav​

CODE001_BACC001_ Bridge_Spa_9_Pre_20230914.wav​

Now we only want 1 clipped audio with the entire task “CODE001_BACC001_ Bridge_Spa_Pre_20230914.wav​” including clinican speech asking the question. Clip from the first question until the last production of the participant.

Guide on how to clip audios

Clipping audios 1. How to clip audios with audacity_2023120




Steps

Process

To Do



1

Look for the assigned Patient

Access the Clipping reports enter to each report to see which samples are remaining to clip. Those reports are a summarized version of the main Connected Speech Smartsheets and the data comunicates automatically (if you fill the reports out the data will appear in the main smartsheets)

20251023 note: filter removed from the reports for acoustic check task, Arely please add again once task is complete

Spanish CS Speech Data Analysis,Catalan CS Speech Data Analysis, or English CS Speech Data Analysis Smartsheet

  1. Make sure the clinician associated with the patient has uploaded the audio to box; this can be found under the column "Audios uploaded and copied to box" and that your name appears in the "RA student" column

  2. Under the column "Enrolled in Study," check type of study





1.1

Access the Transcription Folder that Corresponds with the Patient's Enrollment in the Study

  1. Select the appropriate Transcription Folder based on the patient's enrollment:

    1. Therapy trial or Observational trial


    2. Spanish, Catalan, or English



      (This can be found highlighted in yellow under the "Transcriptions" column)


      Here you will find the audios:



1.2

Download the audio file (or files)

  1. Download the audio file associated with the patient.

  2. The patient code will be the first part of the title (ex: BISE004_BACC003_CS1_Cat_Obs1part1_20230515.wav)



2

Open Audacity

Open Audacity Version 3.4.2

*If audacity is showing an error page when trying to import audio files, follow the instructions using the following link to install the FFmpeg plug-in: https://support.audacityteam.org/basics/installing-ffmpeg



2.1

Open the patient audio file

  1. On audacity, look for the tab "File/Activo," then select "open/abrir" from the drop-down menu




    1. Choose the patient audio file you wish to clip from your downloads

  2. Once you've located the patient audio in Audacity, click "open/abrir"




3

Clipping audio files in Audacity:
Listen to the audio file

  1. Listen to the audio sample by pressing

    on the top left-hand corner.

    1. Listen until you've reached the beginning of a task, then press "Control +B" to clip the audio (Make sure to press both buttons at the same time)

    2. Make sure to check that there is no clinician speech!





3.1

Name the task

1. Use the Naming Convention associated with the task and enter it into the purple box:


***Note that all Non-R01 and R01 patients will have set 3 in addition to Set 1 or Set 2, and Pre-R01s will only have set 3***